update code

.circleci/config.yml

+version: 2.1
+
+# this allows you to use CircleCI's dynamic configuration feature
+setup: true
+
+# the path-filtering orb is required to continue a pipeline based on
+# the path of an updated fileset
+orbs:
+  path-filtering: circleci/path-filtering@0.1.2
+
+workflows:
+  # the always-run workflow is always triggered, regardless of the pipeline parameters.
+  always-run:
+    jobs:
+      # the path-filtering/filter job determines which pipeline
+      # parameters to update.
+      - path-filtering/filter:
+          name: check-updated-files
+          # 3-column, whitespace-delimited mapping. One mapping per
+          # line:
+          # <regex path-to-test> <parameter-to-set> <value-of-pipeline-parameter>
+          mapping: |
+            mmdet/.* lint_only false
+            requirements/.* lint_only false
+            tests/.* lint_only false
+            tools/.* lint_only false
+            configs/.* lint_only false
+            .circleci/.* lint_only false
+          base-revision: dev-3.x
+          # this is the path of the configuration we should trigger once
+          # path filtering and pipeline parameter value updates are
+          # complete. In this case, we are using the parent dynamic
+          # configuration itself.
+          config-path: .circleci/test.yml

.circleci/test.yml

+version: 2.1
+
+# the default pipeline parameters, which will be updated according to
+# the results of the path-filtering orb
+parameters:
+  lint_only:
+    type: boolean
+    default: true
+
+jobs:
+  lint:
+    docker:
+      - image: cimg/python:3.7.4
+    steps:
+      - checkout
+      - run:
+          name: Install pre-commit hook
+          command: |
+            pip install pre-commit
+            pre-commit install
+      - run:
+          name: Linting
+          command: pre-commit run --all-files
+      - run:
+          name: Check docstring coverage
+          command: |
+            pip install interrogate
+            interrogate -v --ignore-init-method --ignore-module --ignore-nested-functions --ignore-magic --ignore-regex "__repr__" --fail-under 85 mmdet
+
+  build_cpu:
+    parameters:
+      # The python version must match available image tags in
+      # https://circleci.com/developer/images/image/cimg/python
+      python:
+        type: string
+      torch:
+        type: string
+      torchvision:
+        type: string
+    docker:
+      - image: cimg/python:<< parameters.python >>
+    resource_class: large
+    steps:
+      - checkout
+      - run:
+          name: Install Libraries
+          command: |
+            sudo apt-get update
+            sudo apt-get install -y ninja-build libglib2.0-0 libsm6 libxrender-dev libxext6 libgl1-mesa-glx libjpeg-dev zlib1g-dev libtinfo-dev libncurses5
+      - run:
+          name: Configure Python & pip
+          command: |
+            pip install --upgrade pip
+            pip install wheel
+      - run:
+          name: Install PyTorch
+          command: |
+            python -V
+            python -m pip install torch==<< parameters.torch >>+cpu torchvision==<< parameters.torchvision >>+cpu -f https://download.pytorch.org/whl/torch_stable.html
+      - when:
+          condition:
+            equal: ["3.9.0", << parameters.python >>]
+          steps:
+            - run: pip install "protobuf <= 3.20.1" && sudo apt-get update && sudo apt-get -y install libprotobuf-dev protobuf-compiler cmake
+            - run: pip install dsdl
+      - run:
+          name: Install mmdet dependencies
+          # numpy may be downgraded after building pycocotools, which causes `ImportError: numpy.core.multiarray failed to import`
+          # force reinstall pycocotools to ensure pycocotools being built under the currenct numpy
+          command: |
+            python -m pip install git+ssh://git@github.com/open-mmlab/mmengine.git@main
+            pip install -U openmim
+            mim install 'mmcv >= 2.0.0rc4'
+            pip install -r requirements/tests.txt -r requirements/optional.txt
+            pip install --force-reinstall pycocotools
+            pip install albumentations>=0.3.2 --no-binary imgaug,albumentations
+            pip install -r requirements/tracking.txt
+            pip install git+https://github.com/cocodataset/panopticapi.git
+            pip install git+https://github.com/JonathonLuiten/TrackEval.git
+      - run:
+          name: Build and install
+          command: |
+            pip install -e .
+      - run:
+          name: Run unittests
+          command: |
+            python -m coverage run --branch --source mmdet -m pytest tests/
+            python -m coverage xml
+            python -m coverage report -m
+
+  build_cuda:
+    parameters:
+      torch:
+        type: string
+      cuda:
+        type: enum
+        enum: ["11.1", "11.7", "11.8"]
+      cudnn:
+        type: integer
+        default: 8
+    machine:
+      image: linux-cuda-11:default
+      # docker_layer_caching: true
+    resource_class: gpu.nvidia.small.multi
+    steps:
+      - checkout
+      - run:
+          # CLoning repos in VM since Docker doesn't have access to the private key
+          name: Clone Repos
+          command: |
+            git clone -b main --depth 1 ssh://git@github.com/open-mmlab/mmengine.git /home/circleci/mmengine
+      - run:
+          name: Install nvidia-container-toolkit and Restart Docker
+          command: |
+            sudo apt-get update
+            sudo apt-get install -y nvidia-container-toolkit
+            sudo systemctl restart docker
+      - run:
+          name: Build Docker image
+          command: |
+            docker build .circleci/docker -t mmdetection:gpu --build-arg PYTORCH=<< parameters.torch >> --build-arg CUDA=<< parameters.cuda >> --build-arg CUDNN=<< parameters.cudnn >>
+            docker run --gpus all -t -d -v /home/circleci/project:/mmdetection -v /home/circleci/mmengine:/mmengine -w /mmdetection --name mmdetection mmdetection:gpu
+            docker exec mmdetection apt-get install -y git
+      - run:
+          name: Install mmdet dependencies
+          command: |
+            docker exec mmdetection pip install -e /mmengine
+            docker exec mmdetection pip install -U openmim
+            docker exec mmdetection mim install 'mmcv >= 2.0.0rc4'
+            docker exec mmdetection pip install -r requirements/tests.txt -r requirements/optional.txt
+            docker exec mmdetection pip install pycocotools
+            docker exec mmdetection pip install albumentations>=0.3.2 --no-binary imgaug,albumentations
+            docker exec mmdetection pip install -r requirements/tracking.txt
+            docker exec mmdetection pip install git+https://github.com/cocodataset/panopticapi.git
+            docker exec mmdetection pip install git+https://github.com/JonathonLuiten/TrackEval.git
+            docker exec mmdetection python -c 'import mmcv; print(mmcv.__version__)'
+      - run:
+          name: Build and install
+          command: |
+            docker exec mmdetection pip install -e .
+      - run:
+          name: Run unittests
+          command: |
+            docker exec mmdetection python -m pytest tests/
+
+workflows:
+  pr_stage_lint:
+    when: << pipeline.parameters.lint_only >>
+    jobs:
+      - lint:
+          name: lint
+          filters:
+            branches:
+              ignore:
+                - dev-3.x
+  pr_stage_test:
+    when:
+      not: << pipeline.parameters.lint_only >>
+    jobs:
+      - lint:
+          name: lint
+          filters:
+            branches:
+              ignore:
+                - dev-3.x
+      - build_cpu:
+          name: minimum_version_cpu
+          torch: 1.8.0
+          torchvision: 0.9.0
+          python: 3.7.16
+          requires:
+            - lint
+      - build_cpu:
+          name: maximum_version_cpu
+          torch: 2.0.0
+          torchvision: 0.15.1
+          python: 3.9.0
+          requires:
+            - minimum_version_cpu
+      - hold:
+          type: approval
+          requires:
+            - maximum_version_cpu
+      - build_cuda:
+          name: mainstream_version_gpu
+          torch: 1.8.1
+          # Use double quotation mark to explicitly specify its type
+          # as string instead of number
+          cuda: "11.1"
+          requires:
+            - hold
+      - build_cuda:
+          name: maximum_version_gpu
+          torch: 2.0.0
+          cuda: "11.7"
+          cudnn: 8
+          requires:
+            - hold
+  merge_stage_test:
+    when:
+      not: << pipeline.parameters.lint_only >>
+    jobs:
+      - build_cuda:
+          name: minimum_version_gpu
+          torch: 1.8.0
+          cuda: "11.1"
+          filters:
+            branches:
+              only:
+                - dev-3.x

.dev_scripts/batch_test_list.py

+# Copyright (c) OpenMMLab. All rights reserved.
+
+# missing wider_face/timm_example/strong_baselines/simple_copy_paste/
+# selfsup_pretrain/seesaw_loss/pascal_voc/openimages/lvis/ld/lad/cityscapes/deepfashion
+
+# yapf: disable
+atss = dict(
+    config='configs/atss/atss_r50_fpn_1x_coco.py',
+    checkpoint='atss_r50_fpn_1x_coco_20200209-985f7bd0.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/atss/atss_r50_fpn_1x_coco/atss_r50_fpn_1x_coco_20200209-985f7bd0.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=39.4),
+)
+autoassign = dict(
+    config='configs/autoassign/autoassign_r50-caffe_fpn_1x_coco.py',
+    checkpoint='auto_assign_r50_fpn_1x_coco_20210413_115540-5e17991f.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/autoassign/auto_assign_r50_fpn_1x_coco/auto_assign_r50_fpn_1x_coco_20210413_115540-5e17991f.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.4),
+)
+carafe = dict(
+    config='configs/carafe/faster-rcnn_r50_fpn-carafe_1x_coco.py',
+    checkpoint='faster_rcnn_r50_fpn_carafe_1x_coco_bbox_mAP-0.386_20200504_175733-385a75b7.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/carafe/faster_rcnn_r50_fpn_carafe_1x_coco/faster_rcnn_r50_fpn_carafe_1x_coco_bbox_mAP-0.386_20200504_175733-385a75b7.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=38.6),
+)
+cascade_rcnn = [
+    dict(
+        config='configs/cascade_rcnn/cascade-rcnn_r50_fpn_1x_coco.py',
+        checkpoint='cascade_rcnn_r50_fpn_1x_coco_20200316-3dc56deb.pth',
+        eval='bbox',
+        url='https://download.openmmlab.com/mmdetection/v2.0/cascade_rcnn/cascade_rcnn_r50_fpn_1x_coco/cascade_rcnn_r50_fpn_1x_coco_20200316-3dc56deb.pth', # noqa
+        metric=dict(bbox_mAP=40.3),
+    ),
+    dict(
+        config='configs/cascade_rcnn/cascade-mask-rcnn_r50_fpn_1x_coco.py',
+        checkpoint='cascade_mask_rcnn_r50_fpn_1x_coco_20200203-9d4dcb24.pth',
+        url='https://download.openmmlab.com/mmdetection/v2.0/cascade_rcnn/cascade_mask_rcnn_r50_fpn_1x_coco/cascade_mask_rcnn_r50_fpn_1x_coco_20200203-9d4dcb24.pth', # noqa
+        eval=['bbox', 'segm'],
+        metric=dict(bbox_mAP=41.2, segm_mAP=35.9),
+    ),
+]
+cascade_rpn = dict(
+    config='configs/cascade_rpn/cascade-rpn_faster-rcnn_r50-caffe_fpn_1x_coco.py', # noqa
+    checkpoint='crpn_faster_rcnn_r50_caffe_fpn_1x_coco-c8283cca.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/cascade_rpn/crpn_faster_rcnn_r50_caffe_fpn_1x_coco/crpn_faster_rcnn_r50_caffe_fpn_1x_coco-c8283cca.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.4),
+)
+centernet = dict(
+    config='configs/centernet/centernet_r18-dcnv2_8xb16-crop512-140e_coco.py',
+    checkpoint='centernet_resnet18_dcnv2_140e_coco_20210702_155131-c8cd631f.pth', # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/centernet/centernet_resnet18_dcnv2_140e_coco/centernet_resnet18_dcnv2_140e_coco_20210702_155131-c8cd631f.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=29.5),
+)
+centripetalnet = dict(
+    config='configs/centripetalnet/centripetalnet_hourglass104_16xb6-crop511-210e-mstest_coco.py',  # noqa
+    checkpoint='centripetalnet_hourglass104_mstest_16x6_210e_coco_20200915_204804-3ccc61e5.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/centripetalnet/centripetalnet_hourglass104_mstest_16x6_210e_coco/centripetalnet_hourglass104_mstest_16x6_210e_coco_20200915_204804-3ccc61e5.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=44.7),
+)
+convnext = dict(
+    config='configs/convnext/cascade-mask-rcnn_convnext-s-p4-w7_fpn_4conv1fc-giou_amp-ms-crop-3x_coco.py', # noqa
+    checkpoint='cascade_mask_rcnn_convnext-s_p4_w7_fpn_giou_4conv1f_fp16_ms-crop_3x_coco_20220510_201004-3d24f5a4.pth', # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/convnext/cascade_mask_rcnn_convnext-s_p4_w7_fpn_giou_4conv1f_fp16_ms-crop_3x_coco/cascade_mask_rcnn_convnext-s_p4_w7_fpn_giou_4conv1f_fp16_ms-crop_3x_coco_20220510_201004-3d24f5a4.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=51.8, segm_mAP=44.8),
+)
+cornernet = dict(
+    config='configs/cornernet/cornernet_hourglass104_8xb6-210e-mstest_coco.py',
+    checkpoint='cornernet_hourglass104_mstest_8x6_210e_coco_20200825_150618-79b44c30.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/cornernet/cornernet_hourglass104_mstest_8x6_210e_coco/cornernet_hourglass104_mstest_8x6_210e_coco_20200825_150618-79b44c30.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=41.2),
+)
+dcn = dict(
+    config='configs/dcn/faster-rcnn_r50-dconv-c3-c5_fpn_1x_coco.py',
+    checkpoint='faster_rcnn_r50_fpn_dconv_c3-c5_1x_coco_20200130-d68aed1e.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/dcn/faster_rcnn_r50_fpn_dconv_c3-c5_1x_coco/faster_rcnn_r50_fpn_dconv_c3-c5_1x_coco_20200130-d68aed1e.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=41.3),
+)
+dcnv2 = dict(
+    config='configs/dcnv2/faster-rcnn_r50_fpn_mdpool_1x_coco.py',
+    checkpoint='faster_rcnn_r50_fpn_mdpool_1x_coco_20200307-c0df27ff.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/dcn/faster_rcnn_r50_fpn_mdpool_1x_coco/faster_rcnn_r50_fpn_mdpool_1x_coco_20200307-c0df27ff.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=38.7),
+)
+ddod = dict(
+    config='configs/ddod/ddod_r50_fpn_1x_coco.py',
+    checkpoint='ddod_r50_fpn_1x_coco_20220523_223737-29b2fc67.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/ddod/ddod_r50_fpn_1x_coco/ddod_r50_fpn_1x_coco_20220523_223737-29b2fc67.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=41.7),
+)
+deformable_detr = dict(
+    config='configs/deformable_detr/deformable-detr_r50_16xb2-50e_coco.py',
+    checkpoint='deformable_detr_r50_16x2_50e_coco_20210419_220030-a12b9512.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/deformable_detr/deformable_detr_r50_16x2_50e_coco/deformable_detr_r50_16x2_50e_coco_20210419_220030-a12b9512.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=44.5),
+)
+detectors = dict(
+    config='configs/detectors/detectors_htc-r50_1x_coco.py',
+    checkpoint='detectors_htc_r50_1x_coco-329b1453.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/detectors/detectors_htc_r50_1x_coco/detectors_htc_r50_1x_coco-329b1453.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=49.1, segm_mAP=42.6),
+)
+detr = dict(
+    config='configs/detr/detr_r50_8xb2-150e_coco.py',
+    checkpoint='detr_r50_8x2_150e_coco_20201130_194835-2c4b8974.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/detr/detr_r50_8x2_150e_coco/detr_r50_8x2_150e_coco_20201130_194835-2c4b8974.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.1),
+)
+double_heads = dict(
+    config='configs/double_heads/dh-faster-rcnn_r50_fpn_1x_coco.py',
+    checkpoint='dh_faster_rcnn_r50_fpn_1x_coco_20200130-586b67df.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/double_heads/dh_faster_rcnn_r50_fpn_1x_coco/dh_faster_rcnn_r50_fpn_1x_coco_20200130-586b67df.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.0),
+)
+dyhead = dict(
+    config='configs/dyhead/atss_r50_fpn_dyhead_1x_coco.py',
+    checkpoint='atss_r50_fpn_dyhead_4x4_1x_coco_20211219_023314-eaa620c6.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/dyhead/atss_r50_fpn_dyhead_4x4_1x_coco/atss_r50_fpn_dyhead_4x4_1x_coco_20211219_023314-eaa620c6.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=43.3),
+)
+dynamic_rcnn = dict(
+    config='configs/dynamic_rcnn/dynamic-rcnn_r50_fpn_1x_coco.py',
+    checkpoint='dynamic_rcnn_r50_fpn_1x-62a3f276.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/dynamic_rcnn/dynamic_rcnn_r50_fpn_1x/dynamic_rcnn_r50_fpn_1x-62a3f276.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=38.9),
+)
+efficientnet = dict(
+    config='configs/efficientnet/retinanet_effb3_fpn_8xb4-crop896-1x_coco.py',
+    checkpoint='retinanet_effb3_fpn_crop896_8x4_1x_coco_20220322_234806-615a0dda.pth', # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/efficientnet/retinanet_effb3_fpn_crop896_8x4_1x_coco/retinanet_effb3_fpn_crop896_8x4_1x_coco_20220322_234806-615a0dda.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.5),
+)
+empirical_attention = dict(
+    config='configs/empirical_attention/faster-rcnn_r50-attn1111_fpn_1x_coco.py',  # noqa
+    checkpoint='faster_rcnn_r50_fpn_attention_1111_1x_coco_20200130-403cccba.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/empirical_attention/faster_rcnn_r50_fpn_attention_1111_1x_coco/faster_rcnn_r50_fpn_attention_1111_1x_coco_20200130-403cccba.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.0),
+)
+faster_rcnn = dict(
+    config='configs/faster_rcnn/faster-rcnn_r50_fpn_1x_coco.py',
+    checkpoint='faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/faster_rcnn/faster_rcnn_r50_fpn_1x_coco/faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=37.4),
+)
+fcos = dict(
+    config='configs/fcos/fcos_r50-caffe_fpn_gn-head-center-normbbox-centeronreg-giou_1x_coco.py',  # noqa
+    checkpoint='fcos_center-normbbox-centeronreg-giou_r50_caffe_fpn_gn-head_1x_coco-0a0d75a8.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/fcos/fcos_center-normbbox-centeronreg-giou_r50_caffe_fpn_gn-head_1x_coco/fcos_center-normbbox-centeronreg-giou_r50_caffe_fpn_gn-head_1x_coco-0a0d75a8.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=38.7),
+)
+foveabox = dict(
+    config='configs/foveabox/fovea_r50_fpn_gn-head-align_4xb4-2x_coco.py',
+    checkpoint='fovea_align_r50_fpn_gn-head_4x4_2x_coco_20200203-8987880d.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/foveabox/fovea_align_r50_fpn_gn-head_4x4_2x_coco/fovea_align_r50_fpn_gn-head_4x4_2x_coco_20200203-8987880d.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=37.9),
+)
+fpg = dict(
+    config='configs/fpg/mask-rcnn_r50_fpg_crop640-50e_coco.py',
+    checkpoint='mask_rcnn_r50_fpg_crop640_50e_coco_20220311_011857-233b8334.pth', # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/fpg/mask_rcnn_r50_fpg_crop640_50e_coco/mask_rcnn_r50_fpg_crop640_50e_coco_20220311_011857-233b8334.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=43.0, segm_mAP=38.1),
+)
+free_anchor = dict(
+    config='configs/free_anchor/freeanchor_r50_fpn_1x_coco.py',
+    checkpoint='retinanet_free_anchor_r50_fpn_1x_coco_20200130-0f67375f.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/free_anchor/retinanet_free_anchor_r50_fpn_1x_coco/retinanet_free_anchor_r50_fpn_1x_coco_20200130-0f67375f.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=38.7),
+)
+fsaf = dict(
+    config='configs/fsaf/fsaf_r50_fpn_1x_coco.py',
+    checkpoint='fsaf_r50_fpn_1x_coco-94ccc51f.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/fsaf/fsaf_r50_fpn_1x_coco/fsaf_r50_fpn_1x_coco-94ccc51f.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=37.4),
+)
+gcnet = dict(
+    config='configs/gcnet/mask-rcnn_r50-syncbn-gcb-r16-c3-c5_fpn_1x_coco.py',  # noqa
+    checkpoint='mask_rcnn_r50_fpn_syncbn-backbone_r16_gcb_c3-c5_1x_coco_20200202-587b99aa.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/gcnet/mask_rcnn_r50_fpn_syncbn-backbone_r16_gcb_c3-c5_1x_coco/mask_rcnn_r50_fpn_syncbn-backbone_r16_gcb_c3-c5_1x_coco_20200202-587b99aa.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=40.4, segm_mAP=36.2),
+)
+gfl = dict(
+    config='configs/gfl/gfl_r50_fpn_1x_coco.py',
+    checkpoint='gfl_r50_fpn_1x_coco_20200629_121244-25944287.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/gfl/gfl_r50_fpn_1x_coco/gfl_r50_fpn_1x_coco_20200629_121244-25944287.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.2),
+)
+ghm = dict(
+    config='configs/ghm/retinanet_r50_fpn_ghm-1x_coco.py',
+    checkpoint='retinanet_ghm_r50_fpn_1x_coco_20200130-a437fda3.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/ghm/retinanet_ghm_r50_fpn_1x_coco/retinanet_ghm_r50_fpn_1x_coco_20200130-a437fda3.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=37.0),
+)
+gn = dict(
+    config='configs/gn/mask-rcnn_r50_fpn_gn-all_2x_coco.py',
+    checkpoint='mask_rcnn_r50_fpn_gn-all_2x_coco_20200206-8eee02a6.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/gn/mask_rcnn_r50_fpn_gn-all_2x_coco/mask_rcnn_r50_fpn_gn-all_2x_coco_20200206-8eee02a6.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=40.1, segm_mAP=36.4),
+)
+gn_ws = dict(
+    config='configs/gn+ws/faster-rcnn_r50_fpn_gn-ws-all_1x_coco.py',
+    checkpoint='faster_rcnn_r50_fpn_gn_ws-all_1x_coco_20200130-613d9fe2.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/faster_rcnn_r50_fpn_gn_ws-all_1x_coco/faster_rcnn_r50_fpn_gn_ws-all_1x_coco_20200130-613d9fe2.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=39.7),
+)
+grid_rcnn = dict(
+    config='configs/grid_rcnn/grid-rcnn_r50_fpn_gn-head_2x_coco.py',
+    checkpoint='grid_rcnn_r50_fpn_gn-head_2x_coco_20200130-6cca8223.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/grid_rcnn/grid_rcnn_r50_fpn_gn-head_2x_coco/grid_rcnn_r50_fpn_gn-head_2x_coco_20200130-6cca8223.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.4),
+)
+groie = dict(
+    config='configs/groie/faste-rcnn_r50_fpn_groie_1x_coco.py',
+    checkpoint='faster_rcnn_r50_fpn_groie_1x_coco_20200604_211715-66ee9516.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/groie/faster_rcnn_r50_fpn_groie_1x_coco/faster_rcnn_r50_fpn_groie_1x_coco_20200604_211715-66ee9516.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=38.3),
+)
+guided_anchoring = dict(
+        config='configs/guided_anchoring/ga-retinanet_r50-caffe_fpn_1x_coco.py',  # noqa
+        checkpoint='ga_retinanet_r50_caffe_fpn_1x_coco_20201020-39581c6f.pth',
+        url='https://download.openmmlab.com/mmdetection/v2.0/guided_anchoring/ga_retinanet_r50_caffe_fpn_1x_coco/ga_retinanet_r50_caffe_fpn_1x_coco_20201020-39581c6f.pth', # noqa
+        eval='bbox',
+        metric=dict(bbox_mAP=36.9),
+    )
+hrnet = dict(
+    config='configs/hrnet/faster-rcnn_hrnetv2p-w18-1x_coco.py',
+    checkpoint='faster_rcnn_hrnetv2p_w18_1x_coco_20200130-56651a6d.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/hrnet/faster_rcnn_hrnetv2p_w18_1x_coco/faster_rcnn_hrnetv2p_w18_1x_coco_20200130-56651a6d.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=36.9),
+)
+htc = dict(
+    config='configs/htc/htc_r50_fpn_1x_coco.py',
+    checkpoint='htc_r50_fpn_1x_coco_20200317-7332cf16.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/htc/htc_r50_fpn_1x_coco/htc_r50_fpn_1x_coco_20200317-7332cf16.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=42.3, segm_mAP=37.4),
+)
+instaboost = dict(
+    config='configs/instaboost/mask-rcnn_r50_fpn_instaboost-4x_coco.py',
+    checkpoint='mask_rcnn_r50_fpn_instaboost_4x_coco_20200307-d025f83a.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/instaboost/mask_rcnn_r50_fpn_instaboost_4x_coco/mask_rcnn_r50_fpn_instaboost_4x_coco_20200307-d025f83a.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=40.6, segm_mAP=36.6),
+)
+libra_rcnn = dict(
+    config='configs/libra_rcnn/libra-faster-rcnn_r50_fpn_1x_coco.py',
+    checkpoint='libra_faster_rcnn_r50_fpn_1x_coco_20200130-3afee3a9.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/libra_rcnn/libra_faster_rcnn_r50_fpn_1x_coco/libra_faster_rcnn_r50_fpn_1x_coco_20200130-3afee3a9.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=38.3),
+)
+mask2former = dict(
+    config='configs/mask2former/mask2former_r50_8xb2-lsj-50e_coco-panoptic.py',
+    checkpoint='mask2former_r50_lsj_8x2_50e_coco-panoptic_20220326_224516-11a44721.pth', # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/mask2former/mask2former_r50_lsj_8x2_50e_coco-panoptic/mask2former_r50_lsj_8x2_50e_coco-panoptic_20220326_224516-11a44721.pth', # noqa
+    eval=['bbox', 'segm', 'PQ'],
+    metric=dict(PQ=51.9, bbox_mAP=44.8, segm_mAP=41.9),
+)
+mask_rcnn = dict(
+    config='configs/mask_rcnn/mask-rcnn_r50_fpn_1x_coco.py',
+    checkpoint='mask_rcnn_r50_fpn_1x_coco_20200205-d4b0c5d6.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/mask_rcnn/mask_rcnn_r50_fpn_1x_coco/mask_rcnn_r50_fpn_1x_coco_20200205-d4b0c5d6.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=38.2, segm_mAP=34.7),
+)
+maskformer = dict(
+    config='configs/maskformer/maskformer_r50_ms-16xb1-75e_coco.py',
+    checkpoint='maskformer_r50_mstrain_16x1_75e_coco_20220221_141956-bc2699cb.pth', # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/maskformer/maskformer_r50_mstrain_16x1_75e_coco/maskformer_r50_mstrain_16x1_75e_coco_20220221_141956-bc2699cb.pth', # noqa
+    eval='PQ',
+    metric=dict(PQ=46.9),
+)
+ms_rcnn = dict(
+    config='configs/ms_rcnn/ms-rcnn_r50-caffe_fpn_1x_coco.py',
+    checkpoint='ms_rcnn_r50_caffe_fpn_1x_coco_20200702_180848-61c9355e.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/ms_rcnn/ms_rcnn_r50_caffe_fpn_1x_coco/ms_rcnn_r50_caffe_fpn_1x_coco_20200702_180848-61c9355e.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=38.2, segm_mAP=36.0),
+)
+nas_fcos = dict(
+    config='configs/nas_fcos/nas-fcos_r50-caffe_fpn_nashead-gn-head_4xb4-1x_coco.py',  # noqa
+    checkpoint='nas_fcos_nashead_r50_caffe_fpn_gn-head_4x4_1x_coco_20200520-1bdba3ce.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/nas_fcos/nas_fcos_nashead_r50_caffe_fpn_gn-head_4x4_1x_coco/nas_fcos_nashead_r50_caffe_fpn_gn-head_4x4_1x_coco_20200520-1bdba3ce.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=39.4),
+)
+nas_fpn = dict(
+    config='configs/nas_fpn/retinanet_r50_nasfpn_crop640-50e_coco.py',
+    checkpoint='retinanet_r50_nasfpn_crop640_50e_coco-0ad1f644.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/nas_fpn/retinanet_r50_nasfpn_crop640_50e_coco/retinanet_r50_nasfpn_crop640_50e_coco-0ad1f644.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.5),
+)
+paa = dict(
+    config='configs/paa/paa_r50_fpn_1x_coco.py',
+    checkpoint='paa_r50_fpn_1x_coco_20200821-936edec3.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/paa/paa_r50_fpn_1x_coco/paa_r50_fpn_1x_coco_20200821-936edec3.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.4),
+)
+pafpn = dict(
+    config='configs/pafpn/faster-rcnn_r50_pafpn_1x_coco.py',
+    checkpoint='faster_rcnn_r50_pafpn_1x_coco_bbox_mAP-0.375_20200503_105836-b7b4b9bd.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/pafpn/faster_rcnn_r50_pafpn_1x_coco/faster_rcnn_r50_pafpn_1x_coco_bbox_mAP-0.375_20200503_105836-b7b4b9bd.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=37.5),
+)
+panoptic_fpn = dict(
+    config='configs/panoptic_fpn/panoptic-fpn_r50_fpn_1x_coco.py',
+    checkpoint='panoptic_fpn_r50_fpn_1x_coco_20210821_101153-9668fd13.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/panoptic_fpn/panoptic_fpn_r50_fpn_1x_coco/panoptic_fpn_r50_fpn_1x_coco_20210821_101153-9668fd13.pth', # noqa
+    eval='PQ',
+    metric=dict(PQ=40.2),
+)
+pisa = dict(
+    config='configs/pisa/faster-rcnn_r50_fpn_pisa_1x_coco.py',
+    checkpoint='pisa_faster_rcnn_r50_fpn_1x_coco-dea93523.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/pisa/pisa_faster_rcnn_r50_fpn_1x_coco/pisa_faster_rcnn_r50_fpn_1x_coco-dea93523.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=38.4),
+)
+point_rend = dict(
+    config='configs/point_rend/point-rend_r50-caffe_fpn_ms-1x_coco.py',
+    checkpoint='point_rend_r50_caffe_fpn_mstrain_1x_coco-1bcb5fb4.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/point_rend/point_rend_r50_caffe_fpn_mstrain_1x_coco/point_rend_r50_caffe_fpn_mstrain_1x_coco-1bcb5fb4.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=38.4, segm_mAP=36.3),
+)
+pvt = dict(
+    config='configs/pvt/retinanet_pvt-s_fpn_1x_coco.py',
+    checkpoint='retinanet_pvt-s_fpn_1x_coco_20210906_142921-b6c94a5b.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/pvt/retinanet_pvt-s_fpn_1x_coco/retinanet_pvt-s_fpn_1x_coco_20210906_142921-b6c94a5b.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=40.4),
+)
+queryinst = dict(
+    config='configs/queryinst/queryinst_r50_fpn_1x_coco.py',
+    checkpoint='queryinst_r50_fpn_1x_coco_20210907_084916-5a8f1998.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/queryinst/queryinst_r50_fpn_1x_coco/queryinst_r50_fpn_1x_coco_20210907_084916-5a8f1998.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=42.0, segm_mAP=37.5),
+)
+regnet = dict(
+    config='configs/regnet/mask-rcnn_regnetx-3.2GF_fpn_1x_coco.py',
+    checkpoint='mask_rcnn_regnetx-3.2GF_fpn_1x_coco_20200520_163141-2a9d1814.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/regnet/mask_rcnn_regnetx-3.2GF_fpn_1x_coco/mask_rcnn_regnetx-3.2GF_fpn_1x_coco_20200520_163141-2a9d1814.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=40.4, segm_mAP=36.7),
+)
+reppoints = dict(
+    config='configs/reppoints/reppoints-moment_r50_fpn_1x_coco.py',
+    checkpoint='reppoints_moment_r50_fpn_1x_coco_20200330-b73db8d1.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/reppoints/reppoints_moment_r50_fpn_1x_coco/reppoints_moment_r50_fpn_1x_coco_20200330-b73db8d1.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=37.0),
+)
+res2net = dict(
+    config='configs/res2net/faster-rcnn_res2net-101_fpn_2x_coco.py',
+    checkpoint='faster_rcnn_r2_101_fpn_2x_coco-175f1da6.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/res2net/faster_rcnn_r2_101_fpn_2x_coco/faster_rcnn_r2_101_fpn_2x_coco-175f1da6.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=43.0),
+)
+resnest = dict(
+    config='configs/resnest/faster-rcnn_s50_fpn_syncbn-backbone+head_ms-range-1x_coco.py',  # noqa
+    checkpoint='faster_rcnn_s50_fpn_syncbn-backbone+head_mstrain-range_1x_coco_20200926_125502-20289c16.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/resnest/faster_rcnn_s50_fpn_syncbn-backbone%2Bhead_mstrain-range_1x_coco/faster_rcnn_s50_fpn_syncbn-backbone%2Bhead_mstrain-range_1x_coco_20200926_125502-20289c16.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=42.0),
+)
+resnet_strikes_back = dict(
+    config='configs/resnet_strikes_back/mask-rcnn_r50-rsb-pre_fpn_1x_coco.py', # noqa
+    checkpoint='mask_rcnn_r50_fpn_rsb-pretrain_1x_coco_20220113_174054-06ce8ba0.pth', # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/resnet_strikes_back/mask_rcnn_r50_fpn_rsb-pretrain_1x_coco/mask_rcnn_r50_fpn_rsb-pretrain_1x_coco_20220113_174054-06ce8ba0.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=41.2, segm_mAP=38.2),
+)
+retinanet = dict(
+    config='configs/retinanet/retinanet_r50_fpn_1x_coco.py',
+    checkpoint='retinanet_r50_fpn_1x_coco_20200130-c2398f9e.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/retinanet/retinanet_r50_fpn_1x_coco/retinanet_r50_fpn_1x_coco_20200130-c2398f9e.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=36.5),
+)
+rpn = dict(
+    config='configs/rpn/rpn_r50_fpn_1x_coco.py',
+    checkpoint='rpn_r50_fpn_1x_coco_20200218-5525fa2e.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/rpn/rpn_r50_fpn_1x_coco/rpn_r50_fpn_1x_coco_20200218-5525fa2e.pth', # noqa
+    eval='proposal_fast',
+    metric=dict(AR_1000=58.2),
+)
+sabl = [
+    dict(
+        config='configs/sabl/sabl-retinanet_r50_fpn_1x_coco.py',
+        checkpoint='sabl_retinanet_r50_fpn_1x_coco-6c54fd4f.pth',
+        url='https://download.openmmlab.com/mmdetection/v2.0/sabl/sabl_retinanet_r50_fpn_1x_coco/sabl_retinanet_r50_fpn_1x_coco-6c54fd4f.pth', # noqa
+        eval='bbox',
+        metric=dict(bbox_mAP=37.7),
+    ),
+    dict(
+        config='configs/sabl/sabl-faster-rcnn_r50_fpn_1x_coco.py',
+        checkpoint='sabl_faster_rcnn_r50_fpn_1x_coco-e867595b.pth',
+        url='https://download.openmmlab.com/mmdetection/v2.0/sabl/sabl_faster_rcnn_r50_fpn_1x_coco/sabl_faster_rcnn_r50_fpn_1x_coco-e867595b.pth', # noqa
+        eval='bbox',
+        metric=dict(bbox_mAP=39.9),
+    ),
+]
+scnet = dict(
+    config='configs/scnet/scnet_r50_fpn_1x_coco.py',
+    checkpoint='scnet_r50_fpn_1x_coco-c3f09857.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/scnet/scnet_r50_fpn_1x_coco/scnet_r50_fpn_1x_coco-c3f09857.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=43.5),
+)
+scratch = dict(
+    config='configs/scratch/mask-rcnn_r50-scratch_fpn_gn-all_6x_coco.py',
+    checkpoint='scratch_mask_rcnn_r50_fpn_gn_6x_bbox_mAP-0.412__segm_mAP-0.374_20200201_193051-1e190a40.pth', # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/scratch/mask_rcnn_r50_fpn_gn-all_scratch_6x_coco/scratch_mask_rcnn_r50_fpn_gn_6x_bbox_mAP-0.412__segm_mAP-0.374_20200201_193051-1e190a40.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=41.2, segm_mAP=37.4),
+)
+solo = dict(
+    config='configs/solo/decoupled-solo_r50_fpn_1x_coco.py',
+    checkpoint='decoupled_solo_r50_fpn_1x_coco_20210820_233348-6337c589.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/solo/decoupled_solo_r50_fpn_1x_coco/decoupled_solo_r50_fpn_1x_coco_20210820_233348-6337c589.pth', # noqa
+    eval='segm',
+    metric=dict(segm_mAP=33.9),
+)
+solov2 = dict(
+    config='configs/solov2/solov2_r50_fpn_1x_coco.py',
+    checkpoint='solov2_r50_fpn_1x_coco_20220512_125858-a357fa23.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/solov2/solov2_r50_fpn_1x_coco/solov2_r50_fpn_1x_coco_20220512_125858-a357fa23.pth', # noqa
+    eval='segm',
+    metric=dict(segm_mAP=34.8),
+)
+sparse_rcnn = dict(
+    config='configs/sparse_rcnn/sparse-rcnn_r50_fpn_1x_coco.py',
+    checkpoint='sparse_rcnn_r50_fpn_1x_coco_20201222_214453-dc79b137.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/sparse_rcnn/sparse_rcnn_r50_fpn_1x_coco/sparse_rcnn_r50_fpn_1x_coco_20201222_214453-dc79b137.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=37.9),
+)
+ssd = [
+    dict(
+        config='configs/ssd/ssd300_coco.py',
+        checkpoint='ssd300_coco_20210803_015428-d231a06e.pth',
+        url='https://download.openmmlab.com/mmdetection/v2.0/ssd/ssd300_coco/ssd300_coco_20210803_015428-d231a06e.pth', # noqa
+        eval='bbox',
+        metric=dict(bbox_mAP=25.5),
+    ),
+    dict(
+        config='configs/ssd/ssdlite_mobilenetv2-scratch_8xb24-600e_coco.py',
+        checkpoint='ssdlite_mobilenetv2_scratch_600e_coco_20210629_110627-974d9307.pth',  # noqa
+        url='https://download.openmmlab.com/mmdetection/v2.0/ssd/ssdlite_mobilenetv2_scratch_600e_coco/ssdlite_mobilenetv2_scratch_600e_coco_20210629_110627-974d9307.pth', # noqa
+        eval='bbox',
+        metric=dict(bbox_mAP=21.3),
+    ),
+]
+swin = dict(
+    config='configs/swin/mask-rcnn_swin-t-p4-w7_fpn_1x_coco.py',
+    checkpoint='mask_rcnn_swin-t-p4-w7_fpn_1x_coco_20210902_120937-9d6b7cfa.pth', # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/swin/mask_rcnn_swin-t-p4-w7_fpn_1x_coco/mask_rcnn_swin-t-p4-w7_fpn_1x_coco_20210902_120937-9d6b7cfa.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=42.7, segm_mAP=39.3),
+)
+tood = dict(
+    config='configs/tood/tood_r50_fpn_1x_coco.py',
+    checkpoint='tood_r50_fpn_1x_coco_20211210_103425-20e20746.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/tood/tood_r50_fpn_1x_coco/tood_r50_fpn_1x_coco_20211210_103425-20e20746.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=42.4),
+)
+tridentnet = dict(
+    config='configs/tridentnet/tridentnet_r50-caffe_1x_coco.py',
+    checkpoint='tridentnet_r50_caffe_1x_coco_20201230_141838-2ec0b530.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/tridentnet/tridentnet_r50_caffe_1x_coco/tridentnet_r50_caffe_1x_coco_20201230_141838-2ec0b530.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=37.6),
+)
+vfnet = dict(
+    config='configs/vfnet/vfnet_r50_fpn_1x_coco.py',
+    checkpoint='vfnet_r50_fpn_1x_coco_20201027-38db6f58.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/vfnet/vfnet_r50_fpn_1x_coco/vfnet_r50_fpn_1x_coco_20201027-38db6f58.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=41.6),
+)
+yolact = dict(
+    config='configs/yolact/yolact_r50_1xb8-55e_coco.py',
+    checkpoint='yolact_r50_1x8_coco_20200908-f38d58df.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/yolact/yolact_r50_1x8_coco/yolact_r50_1x8_coco_20200908-f38d58df.pth', # noqa
+    eval=['bbox', 'segm'],
+    metric=dict(bbox_mAP=31.2, segm_mAP=29.0),
+)
+yolo = dict(
+    config='configs/yolo/yolov3_d53_8xb8-320-273e_coco.py',
+    checkpoint='yolov3_d53_320_273e_coco-421362b6.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/yolo/yolov3_d53_320_273e_coco/yolov3_d53_320_273e_coco-421362b6.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=27.9),
+)
+yolof = dict(
+    config='configs/yolof/yolof_r50-c5_8xb8-1x_coco.py',
+    checkpoint='yolof_r50_c5_8x8_1x_coco_20210425_024427-8e864411.pth',
+    url='https://download.openmmlab.com/mmdetection/v2.0/yolof/yolof_r50_c5_8x8_1x_coco/yolof_r50_c5_8x8_1x_coco_20210425_024427-8e864411.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=37.5),
+)
+yolox = dict(
+    config='configs/yolox/yolox_tiny_8xb8-300e_coco.py',
+    checkpoint='yolox_tiny_8x8_300e_coco_20211124_171234-b4047906.pth',  # noqa
+    url='https://download.openmmlab.com/mmdetection/v2.0/yolox/yolox_tiny_8x8_300e_coco/yolox_tiny_8x8_300e_coco_20211124_171234-b4047906.pth', # noqa
+    eval='bbox',
+    metric=dict(bbox_mAP=31.8),
+)
+# yapf: enable

.dev_scripts/batch_train_list.txt

+configs/albu_example/mask-rcnn_r50_fpn_albu_1x_coco.py
+configs/atss/atss_r50_fpn_1x_coco.py
+configs/autoassign/autoassign_r50-caffe_fpn_1x_coco.py
+configs/carafe/faster-rcnn_r50_fpn-carafe_1x_coco.py
+configs/cascade_rcnn/cascade-rcnn_r50_fpn_1x_coco.py
+configs/cascade_rcnn/cascade-mask-rcnn_r50_fpn_1x_coco.py
+configs/cascade_rpn/cascade-rpn_faster-rcnn_r50-caffe_fpn_1x_coco.py
+configs/centernet/centernet_r18-dcnv2_8xb16-crop512-140e_coco.py
+configs/centernet/centernet-update_r50-caffe_fpn_ms-1x_coco.py
+configs/centripetalnet/centripetalnet_hourglass104_16xb6-crop511-210e-mstest_coco.py
+configs/cornernet/cornernet_hourglass104_8xb6-210e-mstest_coco.py
+configs/convnext/mask-rcnn_convnext-t-p4-w7_fpn_amp-ms-crop-3x_coco.py
+configs/dcn/faster-rcnn_r50-dconv-c3-c5_fpn_1x_coco.py
+configs/dcnv2/faster-rcnn_r50_fpn_mdpool_1x_coco.py
+configs/ddod/ddod_r50_fpn_1x_coco.py
+configs/detectors/detectors_htc-r50_1x_coco.py
+configs/deformable_detr/deformable-detr_r50_16xb2-50e_coco.py
+configs/detr/detr_r50_8xb2-150e_coco.py
+configs/double_heads/dh-faster-rcnn_r50_fpn_1x_coco.py
+configs/dynamic_rcnn/dynamic-rcnn_r50_fpn_1x_coco.py
+configs/dyhead/atss_r50_fpn_dyhead_1x_coco.py
+configs/efficientnet/retinanet_effb3_fpn_8xb4-crop896-1x_coco.py
+configs/empirical_attention/faster-rcnn_r50-attn1111_fpn_1x_coco.py
+configs/faster_rcnn/faster-rcnn_r50_fpn_1x_coco.py
+configs/faster_rcnn/faster-rcnn_r50-caffe-dc5_ms-1x_coco.py
+configs/fcos/fcos_r50-caffe_fpn_gn-head-center-normbbox-centeronreg-giou_1x_coco.py
+configs/foveabox/fovea_r50_fpn_gn-head-align_4xb4-2x_coco.py
+configs/fpg/mask-rcnn_r50_fpg_crop640-50e_coco.py
+configs/free_anchor/freeanchor_r50_fpn_1x_coco.py
+configs/fsaf/fsaf_r50_fpn_1x_coco.py
+configs/gcnet/mask-rcnn_r50-syncbn-gcb-r16-c3-c5_fpn_1x_coco.py
+configs/gfl/gfl_r50_fpn_1x_coco.py
+configs/ghm/retinanet_r50_fpn_ghm-1x_coco.py
+configs/gn/mask-rcnn_r50_fpn_gn-all_2x_coco.py
+configs/gn+ws/faster-rcnn_r50_fpn_gn-ws-all_1x_coco.py
+configs/grid_rcnn/grid-rcnn_r50_fpn_gn-head_2x_coco.py
+configs/groie/faste-rcnn_r50_fpn_groie_1x_coco.py
+configs/guided_anchoring/ga-retinanet_r50-caffe_fpn_1x_coco.py
+configs/hrnet/faster-rcnn_hrnetv2p-w18-1x_coco.py
+configs/htc/htc_r50_fpn_1x_coco.py
+configs/instaboost/mask-rcnn_r50_fpn_instaboost-4x_coco.py
+configs/lad/lad_r50-paa-r101_fpn_2xb8_coco_1x.py
+configs/ld/ld_r18-gflv1-r101_fpn_1x_coco.py
+configs/libra_rcnn/libra-faster-rcnn_r50_fpn_1x_coco.py
+configs/mask2former/mask2former_r50_8xb2-lsj-50e_coco-panoptic.py
+configs/mask_rcnn/mask-rcnn_r50_fpn_1x_coco.py
+configs/maskformer/maskformer_r50_ms-16xb1-75e_coco.py
+configs/ms_rcnn/ms-rcnn_r50-caffe_fpn_1x_coco.py
+configs/nas_fcos/nas-fcos_r50-caffe_fpn_nashead-gn-head_4xb4-1x_coco.py
+configs/nas_fpn/retinanet_r50_nasfpn_crop640-50e_coco.py
+configs/paa/paa_r50_fpn_1x_coco.py
+configs/pafpn/faster-rcnn_r50_pafpn_1x_coco.py
+configs/panoptic_fpn/panoptic-fpn_r50_fpn_1x_coco.py
+configs/pisa/mask-rcnn_r50_fpn_pisa_1x_coco.py
+configs/point_rend/point-rend_r50-caffe_fpn_ms-1x_coco.py
+configs/pvt/retinanet_pvt-t_fpn_1x_coco.py
+configs/queryinst/queryinst_r50_fpn_1x_coco.py
+configs/regnet/retinanet_regnetx-800MF_fpn_1x_coco.py
+configs/reppoints/reppoints-moment_r50_fpn_1x_coco.py
+configs/res2net/faster-rcnn_res2net-101_fpn_2x_coco.py
+configs/resnest/faster-rcnn_s50_fpn_syncbn-backbone+head_ms-range-1x_coco.py
+configs/resnet_strikes_back/retinanet_r50-rsb-pre_fpn_1x_coco.py
+configs/retinanet/retinanet_r50-caffe_fpn_1x_coco.py
+configs/rpn/rpn_r50_fpn_1x_coco.py
+configs/sabl/sabl-retinanet_r50_fpn_1x_coco.py
+configs/scnet/scnet_r50_fpn_1x_coco.py
+configs/scratch/faster-rcnn_r50-scratch_fpn_gn-all_6x_coco.py
+configs/solo/solo_r50_fpn_1x_coco.py
+configs/solov2/solov2_r50_fpn_1x_coco.py
+configs/sparse_rcnn/sparse-rcnn_r50_fpn_1x_coco.py
+configs/ssd/ssd300_coco.py
+configs/ssd/ssdlite_mobilenetv2-scratch_8xb24-600e_coco.py
+configs/swin/mask-rcnn_swin-t-p4-w7_fpn_1x_coco.py
+configs/tood/tood_r50_fpn_1x_coco.py
+'configs/tridentnet/tridentnet_r50-caffe_1x_coco.py
+configs/vfnet/vfnet_r50_fpn_1x_coco.py
+configs/yolact/yolact_r50_8xb8-55e_coco.py
+configs/yolo/yolov3_d53_8xb8-320-273e_coco.py
+configs/yolof/yolof_r50-c5_8xb8-1x_coco.py
+configs/yolox/yolox_tiny_8xb8-300e_coco.py

.dev_scripts/benchmark_filter.py

+# Copyright (c) OpenMMLab. All rights reserved.
+import argparse
+import os
+import os.path as osp
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(description='Filter configs to train')
+    parser.add_argument(
+        '--basic-arch',
+        action='store_true',
+        help='to train models in basic arch')
+    parser.add_argument(
+        '--datasets', action='store_true', help='to train models in dataset')
+    parser.add_argument(
+        '--data-pipeline',
+        action='store_true',
+        help='to train models related to data pipeline, e.g. augmentations')
+    parser.add_argument(
+        '--nn-module',
+        action='store_true',
+        help='to train models related to neural network modules')
+    parser.add_argument(
+        '--model-options',
+        nargs='+',
+        help='custom options to special model benchmark')
+    parser.add_argument(
+        '--out',
+        type=str,
+        default='batch_train_list.txt',
+        help='output path of gathered metrics to be stored')
+    args = parser.parse_args()
+    return args
+
+
+basic_arch_root = [
+    'atss', 'autoassign', 'cascade_rcnn', 'cascade_rpn', 'centripetalnet',
+    'cornernet', 'detectors', 'deformable_detr', 'detr', 'double_heads',
+    'dynamic_rcnn', 'faster_rcnn', 'fcos', 'foveabox', 'fp16', 'free_anchor',
+    'fsaf', 'gfl', 'ghm', 'grid_rcnn', 'guided_anchoring', 'htc', 'ld',
+    'libra_rcnn', 'mask_rcnn', 'ms_rcnn', 'nas_fcos', 'paa', 'pisa',
+    'point_rend', 'reppoints', 'retinanet', 'rpn', 'sabl', 'ssd', 'tridentnet',
+    'vfnet', 'yolact', 'yolo', 'sparse_rcnn', 'scnet', 'yolof', 'centernet'
+]
+
+datasets_root = [
+    'wider_face', 'pascal_voc', 'cityscapes', 'lvis', 'deepfashion'
+]
+
+data_pipeline_root = ['albu_example', 'instaboost']
+
+nn_module_root = [
+    'carafe', 'dcn', 'empirical_attention', 'gcnet', 'gn', 'gn+ws', 'hrnet',
+    'pafpn', 'nas_fpn', 'regnet', 'resnest', 'res2net', 'groie'
+]
+
+benchmark_pool = [
+    'configs/albu_example/mask_rcnn_r50_fpn_albu_1x_coco.py',
+    'configs/atss/atss_r50_fpn_1x_coco.py',
+    'configs/autoassign/autoassign_r50_fpn_8x2_1x_coco.py',
+    'configs/carafe/mask_rcnn_r50_fpn_carafe_1x_coco.py',
+    'configs/cascade_rcnn/cascade_mask_rcnn_r50_fpn_1x_coco.py',
+    'configs/cascade_rpn/crpn_faster_rcnn_r50_caffe_fpn_1x_coco.py',
+    'configs/centernet/centernet_resnet18_dcnv2_140e_coco.py',
+    'configs/centripetalnet/'
+    'centripetalnet_hourglass104_mstest_16x6_210e_coco.py',
+    'configs/cityscapes/mask_rcnn_r50_fpn_1x_cityscapes.py',
+    'configs/cornernet/'
+    'cornernet_hourglass104_mstest_8x6_210e_coco.py',
+    'configs/dcn/mask_rcnn_r50_fpn_mdconv_c3-c5_1x_coco.py',
+    'configs/dcn/faster_rcnn_r50_fpn_dpool_1x_coco.py',
+    'configs/dcn/faster_rcnn_r50_fpn_mdpool_1x_coco.py',
+    'configs/dcn/mask_rcnn_r50_fpn_dconv_c3-c5_1x_coco.py',
+    'configs/deformable_detr/deformable_detr_r50_16x2_50e_coco.py',
+    'configs/detectors/detectors_htc_r50_1x_coco.py',
+    'configs/detr/detr_r50_8x2_150e_coco.py',
+    'configs/double_heads/dh_faster_rcnn_r50_fpn_1x_coco.py',
+    'configs/dynamic_rcnn/dynamic_rcnn_r50_fpn_1x_coco.py',
+    'configs/empirical_attention/faster_rcnn_r50_fpn_attention_1111_dcn_1x_coco.py',  # noqa
+    'configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py',
+    'configs/faster_rcnn/faster_rcnn_r50_fpn_ohem_1x_coco.py',
+    'configs/faster_rcnn/faster_rcnn_r50_caffe_fpn_1x_coco.py',
+    'configs/faster_rcnn/faster_rcnn_r50_caffe_fpn_mstrain_1x_coco.py',
+    'configs/faster_rcnn/faster_rcnn_r50_caffe_dc5_mstrain_1x_coco.py',
+    'configs/fcos/fcos_center_r50_caffe_fpn_gn-head_4x4_1x_coco.py',
+    'configs/foveabox/fovea_align_r50_fpn_gn-head_4x4_2x_coco.py',
+    'configs/retinanet/retinanet_r50_fpn_fp16_1x_coco.py',
+    'configs/mask_rcnn/mask_rcnn_r50_fpn_fp16_1x_coco.py',
+    'configs/free_anchor/retinanet_free_anchor_r50_fpn_1x_coco.py',
+    'configs/fsaf/fsaf_r50_fpn_1x_coco.py',
+    'configs/gcnet/mask_rcnn_r50_fpn_r4_gcb_c3-c5_1x_coco.py',
+    'configs/gfl/gfl_r50_fpn_1x_coco.py',
+    'configs/ghm/retinanet_ghm_r50_fpn_1x_coco.py',
+    'configs/gn/mask_rcnn_r50_fpn_gn-all_2x_coco.py',
+    'configs/gn+ws/mask_rcnn_r50_fpn_gn_ws-all_2x_coco.py',
+    'configs/grid_rcnn/grid_rcnn_r50_fpn_gn-head_2x_coco.py',
+    'configs/groie/faster_rcnn_r50_fpn_groie_1x_coco.py',
+    'configs/guided_anchoring/ga_faster_r50_caffe_fpn_1x_coco.py',
+    'configs/hrnet/mask_rcnn_hrnetv2p_w18_1x_coco.py',
+    'configs/htc/htc_r50_fpn_1x_coco.py',
+    'configs/instaboost/mask_rcnn_r50_fpn_instaboost_4x_coco.py',
+    'configs/ld/ld_r18_gflv1_r101_fpn_coco_1x.py',
+    'configs/libra_rcnn/libra_faster_rcnn_r50_fpn_1x_coco.py',
+    'configs/lvis/mask_rcnn_r50_fpn_sample1e-3_mstrain_1x_lvis_v1.py',
+    'configs/mask_rcnn/mask_rcnn_r50_caffe_fpn_mstrain-poly_1x_coco.py',
+    'configs/ms_rcnn/ms_rcnn_r50_caffe_fpn_1x_coco.py',
+    'configs/nas_fcos/nas_fcos_nashead_r50_caffe_fpn_gn-head_4x4_1x_coco.py',
+    'configs/nas_fpn/retinanet_r50_nasfpn_crop640_50e_coco.py',
+    'configs/paa/paa_r50_fpn_1x_coco.py',
+    'configs/pafpn/faster_rcnn_r50_pafpn_1x_coco.py',
+    'configs/pisa/pisa_mask_rcnn_r50_fpn_1x_coco.py',
+    'configs/point_rend/point_rend_r50_caffe_fpn_mstrain_1x_coco.py',
+    'configs/regnet/mask_rcnn_regnetx-3.2GF_fpn_1x_coco.py',
+    'configs/reppoints/reppoints_moment_r50_fpn_gn-neck+head_1x_coco.py',
+    'configs/res2net/faster_rcnn_r2_101_fpn_2x_coco.py',
+    'configs/resnest/'
+    'mask_rcnn_s50_fpn_syncbn-backbone+head_mstrain_1x_coco.py',
+    'configs/retinanet/retinanet_r50_caffe_fpn_1x_coco.py',
+    'configs/rpn/rpn_r50_fpn_1x_coco.py',
+    'configs/sabl/sabl_retinanet_r50_fpn_1x_coco.py',
+    'configs/ssd/ssd300_coco.py',
+    'configs/tridentnet/tridentnet_r50_caffe_1x_coco.py',
+    'configs/vfnet/vfnet_r50_fpn_1x_coco.py',
+    'configs/yolact/yolact_r50_1x8_coco.py',
+    'configs/yolo/yolov3_d53_320_273e_coco.py',
+    'configs/sparse_rcnn/sparse_rcnn_r50_fpn_1x_coco.py',
+    'configs/scnet/scnet_r50_fpn_1x_coco.py',
+    'configs/yolof/yolof_r50_c5_8x8_1x_coco.py',
+]
+
+
+def main():
+    args = parse_args()
+
+    benchmark_type = []
+    if args.basic_arch:
+        benchmark_type += basic_arch_root
+    if args.datasets:
+        benchmark_type += datasets_root
+    if args.data_pipeline:
+        benchmark_type += data_pipeline_root
+    if args.nn_module:
+        benchmark_type += nn_module_root
+
+    special_model = args.model_options
+    if special_model is not None:
+        benchmark_type += special_model
+
+    config_dpath = 'configs/'
+    benchmark_configs = []
+    for cfg_root in benchmark_type:
+        cfg_dir = osp.join(config_dpath, cfg_root)
+        configs = os.scandir(cfg_dir)
+        for cfg in configs:
+            config_path = osp.join(cfg_dir, cfg.name)
+            if (config_path in benchmark_pool
+                    and config_path not in benchmark_configs):
+                benchmark_configs.append(config_path)
+
+    print(f'Totally found {len(benchmark_configs)} configs to benchmark')
+    with open(args.out, 'w') as f:
+        for config in benchmark_configs:
+            f.write(config + '\n')
+
+
+if __name__ == '__main__':
+    main()

.dev_scripts/benchmark_full_models.txt

+albu_example/mask-rcnn_r50_fpn_albu_1x_coco.py
+atss/atss_r50_fpn_1x_coco.py
+autoassign/autoassign_r50-caffe_fpn_1x_coco.py
+boxinst/boxinst_r50_fpn_ms-90k_coco.py
+carafe/faster-rcnn_r50_fpn-carafe_1x_coco.py
+cascade_rcnn/cascade-rcnn_r50_fpn_1x_coco.py
+cascade_rcnn/cascade-mask-rcnn_r50_fpn_1x_coco.py
+cascade_rpn/cascade-rpn_faster-rcnn_r50-caffe_fpn_1x_coco.py
+centernet/centernet-update_r50-caffe_fpn_ms-1x_coco.py
+centripetalnet/centripetalnet_hourglass104_16xb6-crop511-210e-mstest_coco.py
+condinst/condinst_r50_fpn_ms-poly-90k_coco_instance.py
+conditional_detr/conditional-detr_r50_8xb2-50e_coco.py
+convnext/mask-rcnn_convnext-t-p4-w7_fpn_amp-ms-crop-3x_coco.py
+cornernet/cornernet_hourglass104_8xb6-210e-mstest_coco.py
+dab_detr/dab-detr_r50_8xb2-50e_coco.py
+dcn/mask-rcnn_r50-dconv-c3-c5_fpn_1x_coco.py
+dcnv2/faster-rcnn_r50_fpn_mdpool_1x_coco.py
+ddod/ddod_r50_fpn_1x_coco.py
+deformable_detr/deformable-detr_r50_16xb2-50e_coco.py
+detectors/detectors_htc-r50_1x_coco.py
+detr/detr_r50_8xb2-150e_coco.py
+dino/dino-4scale_r50_8xb2-12e_coco.py
+double_heads/dh-faster-rcnn_r50_fpn_1x_coco.py
+dyhead/atss_r50_fpn_dyhead_1x_coco.py
+dynamic_rcnn/dynamic-rcnn_r50_fpn_1x_coco.py
+efficientnet/retinanet_effb3_fpn_8xb4-crop896-1x_coco.py
+empirical_attention/faster-rcnn_r50-attn0010-dcn_fpn_1x_coco.py
+faster_rcnn/faster-rcnn_r50_fpn_1x_coco.py
+fcos/fcos_r50-caffe_fpn_gn-head-center-normbbox-centeronreg-giou_1x_coco.py
+foveabox/fovea_r50_fpn_gn-head-align_4xb4-2x_coco.py
+fpg/retinanet_r50_fpg_crop640_50e_coco.py
+free_anchor/freeanchor_r50_fpn_1x_coco.py
+fsaf/fsaf_r50_fpn_1x_coco.py
+gcnet/mask-rcnn_r50-gcb-r4-c3-c5_fpn_1x_coco.py
+gfl/gfl_r50_fpn_1x_coco.py
+glip/glip_atss_swin-t_a_fpn_dyhead_pretrain_obj365.py
+ghm/retinanet_r50_fpn_ghm-1x_coco.py
+gn/mask-rcnn_r50_fpn_gn-all_2x_coco.py
+gn+ws/faster-rcnn_r50_fpn_gn-ws-all_1x_coco.py
+grid_rcnn/grid-rcnn_r50_fpn_gn-head_2x_coco.py
+groie/faste-rcnn_r50_fpn_groie_1x_coco.py
+guided_anchoring/ga-faster-rcnn_r50-caffe_fpn_1x_coco.py
+hrnet/htc_hrnetv2p-w18_20e_coco.py
+htc/htc_r50_fpn_1x_coco.py
+instaboost/mask-rcnn_r50_fpn_instaboost-4x_coco.py
+lad/lad_r50-paa-r101_fpn_2xb8_coco_1x.py
+ld/ld_r18-gflv1-r101_fpn_1x_coco.py
+libra_rcnn/libra-faster-rcnn_r50_fpn_1x_coco.py
+lvis/mask-rcnn_r50_fpn_sample1e-3_ms-1x_lvis-v1.py
+mask2former/mask2former_r50_8xb2-lsj-50e_coco.py
+mask2former/mask2former_r50_8xb2-lsj-50e_coco-panoptic.py
+mask_rcnn/mask-rcnn_r50_fpn_1x_coco.py
+maskformer/maskformer_r50_ms-16xb1-75e_coco.py
+ms_rcnn/ms-rcnn_r50-caffe_fpn_1x_coco.py
+nas_fcos/nas-fcos_r50-caffe_fpn_nashead-gn-head_4xb4-1x_coco.py
+nas_fpn/retinanet_r50_nasfpn_crop640-50e_coco.py
+paa/paa_r50_fpn_1x_coco.py
+pafpn/faster-rcnn_r50_pafpn_1x_coco.py
+panoptic_fpn/panoptic-fpn_r50_fpn_1x_coco.py
+pisa/faster-rcnn_r50_fpn_pisa_1x_coco.py
+point_rend/point-rend_r50-caffe_fpn_ms-1x_coco.py
+pvt/retinanet_pvtv2-b0_fpn_1x_coco.py
+queryinst/queryinst_r50_fpn_1x_coco.py
+regnet/mask-rcnn_regnetx-3.2GF_fpn_1x_coco.py
+reppoints/reppoints-moment_r50_fpn-gn_head-gn_1x_coco.py
+res2net/faster-rcnn_res2net-101_fpn_2x_coco.py
+resnest/mask-rcnn_s50_fpn_syncbn-backbone+head_ms-1x_coco.py
+resnet_strikes_back/faster-rcnn_r50-rsb-pre_fpn_1x_coco.py
+retinanet/retinanet_r50_fpn_1x_coco.py
+rpn/rpn_r50_fpn_1x_coco.py
+rtmdet/rtmdet_s_8xb32-300e_coco.py
+rtmdet/rtmdet-ins_s_8xb32-300e_coco.py
+sabl/sabl-retinanet_r50_fpn_1x_coco.py
+scnet/scnet_r50_fpn_1x_coco.py
+scratch/faster-rcnn_r50-scratch_fpn_gn-all_6x_coco.py
+seesaw_loss/mask-rcnn_r50_fpn_seesaw-loss_random-ms-2x_lvis-v1.py
+simple_copy_paste/mask-rcnn_r50_fpn_rpn-2conv_4conv1fc_syncbn-all_32xb2-ssj-scp-90k_coco.py
+soft_teacher/soft-teacher_faster-rcnn_r50-caffe_fpn_180k_semi-0.1-coco.py
+solo/solo_r50_fpn_1x_coco.py
+solov2/solov2_r50_fpn_1x_coco.py
+sparse_rcnn/sparse-rcnn_r50_fpn_1x_coco.py
+ssd/ssd300_coco.py
+swin/mask-rcnn_swin-t-p4-w7_fpn_1x_coco.py
+tood/tood_r50_fpn_1x_coco.py
+tridentnet/tridentnet_r50-caffe_1x_coco.py
+vfnet/vfnet_r50_fpn_1x_coco.py
+yolact/yolact_r50_8xb8-55e_coco.py
+yolo/yolov3_d53_8xb8-320-273e_coco.py
+yolof/yolof_r50-c5_8xb8-1x_coco.py
+yolox/yolox_s_8xb8-300e_coco.py
+deepsort/deepsort_faster-rcnn_r50_fpn_8xb2-4e_mot17halftrain_test-mot17halfval.py
+mask2former_vis/mask2former_r50_8xb2-8e_youtubevis2021.py
+masktrack_rcnn/masktrack-rcnn_mask-rcnn_r50_fpn_8xb1-12e_youtubevis2021.py
+ocsort/ocsort_yolox_x_8xb4-amp-80e_crowdhuman-mot17halftrain_test-mot17halfval.py
+qdtrack/qdtrack_faster-rcnn_r50_fpn_8xb2-4e_mot17halftrain_test-mot17halfval.py
+strongsort/strongsort_yolox_x_8xb4-80e_crowdhuman-mot17halftrain_test-mot17halfval.py

.dev_scripts/benchmark_inference_fps.py

+# Copyright (c) OpenMMLab. All rights reserved.
+import argparse
+import os
+import os.path as osp
+
+from mmengine.config import Config, DictAction
+from mmengine.dist import init_dist
+from mmengine.fileio import dump
+from mmengine.utils import mkdir_or_exist
+from terminaltables import GithubFlavoredMarkdownTable
+
+from tools.analysis_tools.benchmark import repeat_measure_inference_speed
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(
+        description='MMDet benchmark a model of FPS')
+    parser.add_argument('config', help='test config file path')
+    parser.add_argument('checkpoint_root', help='Checkpoint file root path')
+    parser.add_argument(
+        '--round-num',
+        type=int,
+        default=1,
+        help='round a number to a given precision in decimal digits')
+    parser.add_argument(
+        '--repeat-num',
+        type=int,
+        default=1,
+        help='number of repeat times of measurement for averaging the results')
+    parser.add_argument(
+        '--out', type=str, help='output path of gathered fps to be stored')
+    parser.add_argument(
+        '--max-iter', type=int, default=2000, help='num of max iter')
+    parser.add_argument(
+        '--log-interval', type=int, default=50, help='interval of logging')
+    parser.add_argument(
+        '--fuse-conv-bn',
+        action='store_true',
+        help='Whether to fuse conv and bn, this will slightly increase'
+        'the inference speed')
+    parser.add_argument(
+        '--cfg-options',
+        nargs='+',
+        action=DictAction,
+        help='override some settings in the used config, the key-value pair '
+        'in xxx=yyy format will be merged into config file. If the value to '
+        'be overwritten is a list, it should be like key="[a,b]" or key=a,b '
+        'It also allows nested list/tuple values, e.g. key="[(a,b),(c,d)]" '
+        'Note that the quotation marks are necessary and that no white space '
+        'is allowed.')
+    parser.add_argument(
+        '--launcher',
+        choices=['none', 'pytorch', 'slurm', 'mpi'],
+        default='none',
+        help='job launcher')
+    parser.add_argument('--local_rank', type=int, default=0)
+    args = parser.parse_args()
+    if 'LOCAL_RANK' not in os.environ:
+        os.environ['LOCAL_RANK'] = str(args.local_rank)
+    return args
+
+
+def results2markdown(result_dict):
+    table_data = []
+    is_multiple_results = False
+    for cfg_name, value in result_dict.items():
+        name = cfg_name.replace('configs/', '')
+        fps = value['fps']
+        ms_times_pre_image = value['ms_times_pre_image']
+        if isinstance(fps, list):
+            is_multiple_results = True
+            mean_fps = value['mean_fps']
+            mean_times_pre_image = value['mean_times_pre_image']
+            fps_str = ','.join([str(s) for s in fps])
+            ms_times_pre_image_str = ','.join(
+                [str(s) for s in ms_times_pre_image])
+            table_data.append([
+                name, fps_str, mean_fps, ms_times_pre_image_str,
+                mean_times_pre_image
+            ])
+        else:
+            table_data.append([name, fps, ms_times_pre_image])
+
+    if is_multiple_results:
+        table_data.insert(0, [
+            'model', 'fps', 'mean_fps', 'times_pre_image(ms)',
+            'mean_times_pre_image(ms)'
+        ])
+
+    else:
+        table_data.insert(0, ['model', 'fps', 'times_pre_image(ms)'])
+    table = GithubFlavoredMarkdownTable(table_data)
+    print(table.table, flush=True)
+
+
+if __name__ == '__main__':
+    args = parse_args()
+    assert args.round_num >= 0
+    assert args.repeat_num >= 1
+
+    config = Config.fromfile(args.config)
+
+    if args.launcher == 'none':
+        raise NotImplementedError('Only supports distributed mode')
+    else:
+        init_dist(args.launcher)
+
+    result_dict = {}
+    for model_key in config:
+        model_infos = config[model_key]
+        if not isinstance(model_infos, list):
+            model_infos = [model_infos]
+        for model_info in model_infos:
+            record_metrics = model_info['metric']
+            cfg_path = model_info['config'].strip()
+            cfg = Config.fromfile(cfg_path)
+            checkpoint = osp.join(args.checkpoint_root,
+                                  model_info['checkpoint'].strip())
+            try:
+                fps = repeat_measure_inference_speed(cfg, checkpoint,
+                                                     args.max_iter,
+                                                     args.log_interval,
+                                                     args.fuse_conv_bn,
+                                                     args.repeat_num)
+                if args.repeat_num > 1:
+                    fps_list = [round(fps_, args.round_num) for fps_ in fps]
+                    times_pre_image_list = [
+                        round(1000 / fps_, args.round_num) for fps_ in fps
+                    ]
+                    mean_fps = round(
+                        sum(fps_list) / len(fps_list), args.round_num)
+                    mean_times_pre_image = round(
+                        sum(times_pre_image_list) / len(times_pre_image_list),
+                        args.round_num)
+                    print(
+                        f'{cfg_path} '
+                        f'Overall fps: {fps_list}[{mean_fps}] img / s, '
+                        f'times per image: '
+                        f'{times_pre_image_list}[{mean_times_pre_image}] '
+                        f'ms / img',
+                        flush=True)
+                    result_dict[cfg_path] = dict(
+                        fps=fps_list,
+                        mean_fps=mean_fps,
+                        ms_times_pre_image=times_pre_image_list,
+                        mean_times_pre_image=mean_times_pre_image)
+                else:
+                    print(
+                        f'{cfg_path} fps : {fps:.{args.round_num}f} img / s, '
+                        f'times per image: {1000 / fps:.{args.round_num}f} '
+                        f'ms / img',
+                        flush=True)
+                    result_dict[cfg_path] = dict(
+                        fps=round(fps, args.round_num),
+                        ms_times_pre_image=round(1000 / fps, args.round_num))
+            except Exception as e:
+                print(f'{cfg_path} error: {repr(e)}')
+                if args.repeat_num > 1:
+                    result_dict[cfg_path] = dict(
+                        fps=[0],
+                        mean_fps=0,
+                        ms_times_pre_image=[0],
+                        mean_times_pre_image=0)
+                else:
+                    result_dict[cfg_path] = dict(fps=0, ms_times_pre_image=0)
+
+    if args.out:
+        mkdir_or_exist(args.out)
+        dump(result_dict, osp.join(args.out, 'batch_inference_fps.json'))
+
+    results2markdown(result_dict)

.dev_scripts/benchmark_options.py

+# Copyright (c) OpenMMLab. All rights reserved.
+
+third_part_libs = [
+    'pip install -r ../requirements/albu.txt',
+    'pip install instaboostfast',
+    'pip install git+https://github.com/cocodataset/panopticapi.git',
+    'pip install timm',
+    'pip install mmpretrain',
+    'pip install git+https://github.com/lvis-dataset/lvis-api.git',
+    'pip install -r ../requirements/multimodal.txt',
+    'pip install -r ../requirements/tracking.txt',
+    'pip install git+https://github.com/JonathonLuiten/TrackEval.git',
+]
+
+default_floating_range = 0.5
+model_floating_ranges = {'atss/atss_r50_fpn_1x_coco.py': 0.3}

.dev_scripts/benchmark_test.py

+# Copyright (c) OpenMMLab. All rights reserved.
+import logging
+import os
+import os.path as osp
+from argparse import ArgumentParser
+
+from mmengine.config import Config, DictAction
+from mmengine.logging import MMLogger
+from mmengine.registry import RUNNERS
+from mmengine.runner import Runner
+
+from mmdet.testing import replace_to_ceph
+from mmdet.utils import register_all_modules, replace_cfg_vals
+
+
+def parse_args():
+    parser = ArgumentParser()
+    parser.add_argument('config', help='test config file path')
+    parser.add_argument('checkpoint_root', help='Checkpoint file root path')
+    parser.add_argument('--work-dir', help='the dir to save logs')
+    parser.add_argument('--ceph', action='store_true')
+    parser.add_argument(
+        '--cfg-options',
+        nargs='+',
+        action=DictAction,
+        help='override some settings in the used config, the key-value pair '
+        'in xxx=yyy format will be merged into config file. If the value to '
+        'be overwritten is a list, it should be like key="[a,b]" or key=a,b '
+        'It also allows nested list/tuple values, e.g. key="[(a,b),(c,d)]" '
+        'Note that the quotation marks are necessary and that no white space '
+        'is allowed.')
+    parser.add_argument(
+        '--launcher',
+        choices=['none', 'pytorch', 'slurm', 'mpi'],
+        default='none',
+        help='job launcher')
+    parser.add_argument('--local_rank', type=int, default=0)
+    args = parser.parse_args()
+    if 'LOCAL_RANK' not in os.environ:
+        os.environ['LOCAL_RANK'] = str(args.local_rank)
+    args = parser.parse_args()
+    return args
+
+
+# TODO: Need to refactor test.py so that it can be reused.
+def fast_test_model(config_name, checkpoint, args, logger=None):
+    cfg = Config.fromfile(config_name)
+    cfg = replace_cfg_vals(cfg)
+    cfg.launcher = args.launcher
+    if args.cfg_options is not None:
+        cfg.merge_from_dict(args.cfg_options)
+
+    # work_dir is determined in this priority: CLI > segment in file > filename
+    if args.work_dir is not None:
+        # update configs according to CLI args if args.work_dir is not None
+        cfg.work_dir = osp.join(args.work_dir,
+                                osp.splitext(osp.basename(config_name))[0])
+    elif cfg.get('work_dir', None) is None:
+        # use config filename as default work_dir if cfg.work_dir is None
+        cfg.work_dir = osp.join('./work_dirs',
+                                osp.splitext(osp.basename(config_name))[0])
+
+    if args.ceph:
+        replace_to_ceph(cfg)
+
+    cfg.load_from = checkpoint
+
+    # TODO: temporary plan
+    if 'visualizer' in cfg:
+        if 'name' in cfg.visualizer:
+            del cfg.visualizer.name
+
+    # build the runner from config
+    if 'runner_type' not in cfg:
+        # build the default runner
+        runner = Runner.from_cfg(cfg)
+    else:
+        # build customized runner from the registry
+        # if 'runner_type' is set in the cfg
+        runner = RUNNERS.build(cfg)
+
+    runner.test()
+
+
+# Sample test whether the inference code is correct
+def main(args):
+    # register all modules in mmdet into the registries
+    register_all_modules(init_default_scope=False)
+
+    config = Config.fromfile(args.config)
+
+    # test all model
+    logger = MMLogger.get_instance(
+        name='MMLogger',
+        log_file='benchmark_test.log',
+        log_level=logging.ERROR)
+
+    for model_key in config:
+        model_infos = config[model_key]
+        if not isinstance(model_infos, list):
+            model_infos = [model_infos]
+        for model_info in model_infos:
+            print('processing: ', model_info['config'], flush=True)
+            config_name = model_info['config'].strip()
+            checkpoint = osp.join(args.checkpoint_root,
+                                  model_info['checkpoint'].strip())
+            try:
+                fast_test_model(config_name, checkpoint, args, logger)
+            except Exception as e:
+                logger.error(f'{config_name} " : {repr(e)}')
+
+
+if __name__ == '__main__':
+    args = parse_args()
+    main(args)

.dev_scripts/benchmark_test_image.py

+# Copyright (c) OpenMMLab. All rights reserved.
+import logging
+import os.path as osp
+from argparse import ArgumentParser
+
+import mmcv
+from mmengine.config import Config
+from mmengine.logging import MMLogger
+from mmengine.utils import mkdir_or_exist
+
+from mmdet.apis import inference_detector, init_detector
+from mmdet.registry import VISUALIZERS
+from mmdet.utils import register_all_modules
+
+
+def parse_args():
+    parser = ArgumentParser()
+    parser.add_argument('config', help='test config file path')
+    parser.add_argument('checkpoint_root', help='Checkpoint file root path')
+    parser.add_argument('--img', default='demo/demo.jpg', help='Image file')
+    parser.add_argument('--aug', action='store_true', help='aug test')
+    parser.add_argument('--model-name', help='model name to inference')
+    parser.add_argument('--show', action='store_true', help='show results')
+    parser.add_argument('--out-dir', default=None, help='Dir to output file')
+    parser.add_argument(
+        '--wait-time',
+        type=float,
+        default=1,
+        help='the interval of show (s), 0 is block')
+    parser.add_argument(
+        '--device', default='cuda:0', help='Device used for inference')
+    parser.add_argument(
+        '--palette',
+        default='coco',
+        choices=['coco', 'voc', 'citys', 'random'],
+        help='Color palette used for visualization')
+    parser.add_argument(
+        '--score-thr', type=float, default=0.3, help='bbox score threshold')
+    args = parser.parse_args()
+    return args
+
+
+def inference_model(config_name, checkpoint, visualizer, args, logger=None):
+    cfg = Config.fromfile(config_name)
+    if args.aug:
+        raise NotImplementedError()
+
+    model = init_detector(
+        cfg, checkpoint, palette=args.palette, device=args.device)
+    visualizer.dataset_meta = model.dataset_meta
+
+    # test a single image
+    result = inference_detector(model, args.img)
+
+    # show the results
+    if args.show or args.out_dir is not None:
+        img = mmcv.imread(args.img)
+        img = mmcv.imconvert(img, 'bgr', 'rgb')
+        out_file = None
+        if args.out_dir is not None:
+            out_dir = args.out_dir
+            mkdir_or_exist(out_dir)
+
+            out_file = osp.join(
+                out_dir,
+                config_name.split('/')[-1].replace('py', 'jpg'))
+
+        visualizer.add_datasample(
+            'result',
+            img,
+            data_sample=result,
+            draw_gt=False,
+            show=args.show,
+            wait_time=args.wait_time,
+            out_file=out_file,
+            pred_score_thr=args.score_thr)
+
+    return result
+
+
+# Sample test whether the inference code is correct
+def main(args):
+    # register all modules in mmdet into the registries
+    register_all_modules()
+
+    config = Config.fromfile(args.config)
+
+    # init visualizer
+    visualizer_cfg = dict(type='DetLocalVisualizer', name='visualizer')
+    visualizer = VISUALIZERS.build(visualizer_cfg)
+
+    # test single model
+    if args.model_name:
+        if args.model_name in config:
+            model_infos = config[args.model_name]
+            if not isinstance(model_infos, list):
+                model_infos = [model_infos]
+            model_info = model_infos[0]
+            config_name = model_info['config'].strip()
+            print(f'processing: {config_name}', flush=True)
+            checkpoint = osp.join(args.checkpoint_root,
+                                  model_info['checkpoint'].strip())
+            # build the model from a config file and a checkpoint file
+            inference_model(config_name, checkpoint, visualizer, args)
+            return
+        else:
+            raise RuntimeError('model name input error.')
+
+    # test all model
+    logger = MMLogger.get_instance(
+        name='MMLogger',
+        log_file='benchmark_test_image.log',
+        log_level=logging.ERROR)
+
+    for model_key in config:
+        model_infos = config[model_key]
+        if not isinstance(model_infos, list):
+            model_infos = [model_infos]
+        for model_info in model_infos:
+            print('processing: ', model_info['config'], flush=True)
+            config_name = model_info['config'].strip()
+            checkpoint = osp.join(args.checkpoint_root,
+                                  model_info['checkpoint'].strip())
+            try:
+                # build the model from a config file and a checkpoint file
+                inference_model(config_name, checkpoint, visualizer, args,
+                                logger)
+            except Exception as e:
+                logger.error(f'{config_name} " : {repr(e)}')
+
+
+if __name__ == '__main__':
+    args = parse_args()
+    main(args)

.dev_scripts/benchmark_train.py

+# Copyright (c) OpenMMLab. All rights reserved.
+import logging
+import os
+import os.path as osp
+from argparse import ArgumentParser
+
+from mmengine.config import Config, DictAction
+from mmengine.logging import MMLogger, print_log
+from mmengine.registry import RUNNERS
+from mmengine.runner import Runner
+
+from mmdet.testing import replace_to_ceph
+from mmdet.utils import register_all_modules, replace_cfg_vals
+
+
+def parse_args():
+    parser = ArgumentParser()
+    parser.add_argument('config', help='test config file path')
+    parser.add_argument('--work-dir', help='the dir to save logs and models')
+    parser.add_argument('--ceph', action='store_true')
+    parser.add_argument('--save-ckpt', action='store_true')
+    parser.add_argument(
+        '--amp',
+        action='store_true',
+        default=False,
+        help='enable automatic-mixed-precision training')
+    parser.add_argument(
+        '--auto-scale-lr',
+        action='store_true',
+        help='enable automatically scaling LR.')
+    parser.add_argument(
+        '--resume',
+        action='store_true',
+        help='resume from the latest checkpoint in the work_dir automatically')
+    parser.add_argument(
+        '--cfg-options',
+        nargs='+',
+        action=DictAction,
+        help='override some settings in the used config, the key-value pair '
+        'in xxx=yyy format will be merged into config file. If the value to '
+        'be overwritten is a list, it should be like key="[a,b]" or key=a,b '
+        'It also allows nested list/tuple values, e.g. key="[(a,b),(c,d)]" '
+        'Note that the quotation marks are necessary and that no white space '
+        'is allowed.')
+    parser.add_argument(
+        '--launcher',
+        choices=['none', 'pytorch', 'slurm', 'mpi'],
+        default='none',
+        help='job launcher')
+    parser.add_argument('--local_rank', type=int, default=0)
+    args = parser.parse_args()
+    if 'LOCAL_RANK' not in os.environ:
+        os.environ['LOCAL_RANK'] = str(args.local_rank)
+    args = parser.parse_args()
+    return args
+
+
+# TODO: Need to refactor train.py so that it can be reused.
+def fast_train_model(config_name, args, logger=None):
+    cfg = Config.fromfile(config_name)
+    cfg = replace_cfg_vals(cfg)
+    cfg.launcher = args.launcher
+    if args.cfg_options is not None:
+        cfg.merge_from_dict(args.cfg_options)
+
+    # work_dir is determined in this priority: CLI > segment in file > filename
+    if args.work_dir is not None:
+        # update configs according to CLI args if args.work_dir is not None
+        cfg.work_dir = osp.join(args.work_dir,
+                                osp.splitext(osp.basename(config_name))[0])
+    elif cfg.get('work_dir', None) is None:
+        # use config filename as default work_dir if cfg.work_dir is None
+        cfg.work_dir = osp.join('./work_dirs',
+                                osp.splitext(osp.basename(config_name))[0])
+
+    ckpt_hook = cfg.default_hooks.checkpoint
+    by_epoch = ckpt_hook.get('by_epoch', True)
+    fast_stop_hook = dict(type='FastStopTrainingHook')
+    fast_stop_hook['by_epoch'] = by_epoch
+    if args.save_ckpt:
+        if by_epoch:
+            interval = 1
+            stop_iter_or_epoch = 2
+        else:
+            interval = 4
+            stop_iter_or_epoch = 10
+        fast_stop_hook['stop_iter_or_epoch'] = stop_iter_or_epoch
+        fast_stop_hook['save_ckpt'] = True
+        ckpt_hook.interval = interval
+
+    if 'custom_hooks' in cfg:
+        cfg.custom_hooks.append(fast_stop_hook)
+    else:
+        custom_hooks = [fast_stop_hook]
+        cfg.custom_hooks = custom_hooks
+
+    # TODO: temporary plan
+    if 'visualizer' in cfg:
+        if 'name' in cfg.visualizer:
+            del cfg.visualizer.name
+
+    # enable automatic-mixed-precision training
+    if args.amp is True:
+        optim_wrapper = cfg.optim_wrapper.type
+        if optim_wrapper == 'AmpOptimWrapper':
+            print_log(
+                'AMP training is already enabled in your config.',
+                logger='current',
+                level=logging.WARNING)
+        else:
+            assert optim_wrapper == 'OptimWrapper', (
+                '`--amp` is only supported when the optimizer wrapper type is '
+                f'`OptimWrapper` but got {optim_wrapper}.')
+            cfg.optim_wrapper.type = 'AmpOptimWrapper'
+            cfg.optim_wrapper.loss_scale = 'dynamic'
+
+    # enable automatically scaling LR
+    if args.auto_scale_lr:
+        if 'auto_scale_lr' in cfg and \
+                'enable' in cfg.auto_scale_lr and \
+                'base_batch_size' in cfg.auto_scale_lr:
+            cfg.auto_scale_lr.enable = True
+        else:
+            raise RuntimeError('Can not find "auto_scale_lr" or '
+                               '"auto_scale_lr.enable" or '
+                               '"auto_scale_lr.base_batch_size" in your'
+                               ' configuration file.')
+
+    if args.ceph:
+        replace_to_ceph(cfg)
+
+    cfg.resume = args.resume
+
+    # build the runner from config
+    if 'runner_type' not in cfg:
+        # build the default runner
+        runner = Runner.from_cfg(cfg)
+    else:
+        # build customized runner from the registry
+        # if 'runner_type' is set in the cfg
+        runner = RUNNERS.build(cfg)
+
+    runner.train()
+
+
+# Sample test whether the train code is correct
+def main(args):
+    # register all modules in mmdet into the registries
+    register_all_modules(init_default_scope=False)
+
+    config = Config.fromfile(args.config)
+
+    # test all model
+    logger = MMLogger.get_instance(
+        name='MMLogger',
+        log_file='benchmark_train.log',
+        log_level=logging.ERROR)
+
+    for model_key in config:
+        model_infos = config[model_key]
+        if not isinstance(model_infos, list):
+            model_infos = [model_infos]
+        for model_info in model_infos:
+            print('processing: ', model_info['config'], flush=True)
+            config_name = model_info['config'].strip()
+            try:
+                fast_train_model(config_name, args, logger)
+            except RuntimeError as e:
+                # quick exit is the normal exit message
+                if 'quick exit' not in repr(e):
+                    logger.error(f'{config_name} " : {repr(e)}')
+            except Exception as e:
+                logger.error(f'{config_name} " : {repr(e)}')
+
+
+if __name__ == '__main__':
+    args = parse_args()
+    main(args)

.dev_scripts/benchmark_train_models.txt

+atss/atss_r50_fpn_1x_coco.py
+faster_rcnn/faster-rcnn_r50_fpn_1x_coco.py
+mask_rcnn/mask-rcnn_r50_fpn_1x_coco.py
+cascade_rcnn/cascade-mask-rcnn_r50_fpn_1x_coco.py
+configs/grounding_dino/grounding_dino_swin-t_finetune_16xb2_1x_coco.py
+configs/glip/glip_atss_swin-t_a_fpn_dyhead_16xb2_ms-2x_funtune_coco.py
+configs/ddq/ddq-detr-4scale_r50_8xb2-12e_coco.py
+panoptic_fpn/panoptic-fpn_r50_fpn_1x_coco.py
+retinanet/retinanet_r50_fpn_1x_coco.py
+rtmdet/rtmdet_s_8xb32-300e_coco.py
+rtmdet/rtmdet-ins_s_8xb32-300e_coco.py
+fcos/fcos_r50-caffe_fpn_gn-head-center-normbbox-centeronreg-giou_1x_coco.py
+centernet/centernet-update_r50-caffe_fpn_ms-1x_coco.py
+dino/dino-4scale_r50_8xb2-12e_coco.py
+htc/htc_r50_fpn_1x_coco.py
+mask2former/mask2former_r50_8xb2-lsj-50e_coco-panoptic.py
+swin/mask-rcnn_swin-t-p4-w7_fpn_1x_coco.py
+condinst/condinst_r50_fpn_ms-poly-90k_coco_instance.py
+lvis/mask-rcnn_r50_fpn_sample1e-3_ms-1x_lvis-v1.py
+mask2former_vis/mask2former_r50_8xb2-8e_youtubevis2021.py
+masktrack_rcnn/masktrack-rcnn_mask-rcnn_r50_fpn_8xb1-12e_youtubevis2021.py
+qdtrack/qdtrack_faster-rcnn_r50_fpn_8xb2-4e_mot17halftrain_test-mot17halfval.py

.dev_scripts/benchmark_valid_flops.py

+import logging
+import re
+import tempfile
+from argparse import ArgumentParser
+from collections import OrderedDict
+from functools import partial
+from pathlib import Path
+
+import numpy as np
+import pandas as pd
+import torch
+from mmengine import Config, DictAction
+from mmengine.analysis import get_model_complexity_info
+from mmengine.analysis.print_helper import _format_size
+from mmengine.fileio import FileClient
+from mmengine.logging import MMLogger
+from mmengine.model import revert_sync_batchnorm
+from mmengine.runner import Runner
+from modelindex.load_model_index import load
+from rich.console import Console
+from rich.table import Table
+from rich.text import Text
+from tqdm import tqdm
+
+from mmdet.registry import MODELS
+from mmdet.utils import register_all_modules
+
+console = Console()
+MMDET_ROOT = Path(__file__).absolute().parents[1]
+
+
+def parse_args():
+    parser = ArgumentParser(description='Valid all models in model-index.yml')
+    parser.add_argument(
+        '--shape',
+        type=int,
+        nargs='+',
+        default=[1280, 800],
+        help='input image size')
+    parser.add_argument(
+        '--checkpoint_root',
+        help='Checkpoint file root path. If set, load checkpoint before test.')
+    parser.add_argument('--img', default='demo/demo.jpg', help='Image file')
+    parser.add_argument('--models', nargs='+', help='models name to inference')
+    parser.add_argument(
+        '--batch-size',
+        type=int,
+        default=1,
+        help='The batch size during the inference.')
+    parser.add_argument(
+        '--flops', action='store_true', help='Get Flops and Params of models')
+    parser.add_argument(
+        '--flops-str',
+        action='store_true',
+        help='Output FLOPs and params counts in a string form.')
+    parser.add_argument(
+        '--cfg-options',
+        nargs='+',
+        action=DictAction,
+        help='override some settings in the used config, the key-value pair '
+        'in xxx=yyy format will be merged into config file. If the value to '
+        'be overwritten is a list, it should be like key="[a,b]" or key=a,b '
+        'It also allows nested list/tuple values, e.g. key="[(a,b),(c,d)]" '
+        'Note that the quotation marks are necessary and that no white space '
+        'is allowed.')
+    parser.add_argument(
+        '--size_divisor',
+        type=int,
+        default=32,
+        help='Pad the input image, the minimum size that is divisible '
+        'by size_divisor, -1 means do not pad the image.')
+    args = parser.parse_args()
+    return args
+
+
+def inference(config_file, checkpoint, work_dir, args, exp_name):
+    logger = MMLogger.get_instance(name='MMLogger')
+    logger.warning('if you want test flops, please make sure torch>=1.12')
+    cfg = Config.fromfile(config_file)
+    cfg.work_dir = work_dir
+    cfg.load_from = checkpoint
+    cfg.log_level = 'WARN'
+    cfg.experiment_name = exp_name
+    if args.cfg_options is not None:
+        cfg.merge_from_dict(args.cfg_options)
+
+    # forward the model
+    result = {'model': config_file.stem}
+
+    if args.flops:
+
+        if len(args.shape) == 1:
+            h = w = args.shape[0]
+        elif len(args.shape) == 2:
+            h, w = args.shape
+        else:
+            raise ValueError('invalid input shape')
+        divisor = args.size_divisor
+        if divisor > 0:
+            h = int(np.ceil(h / divisor)) * divisor
+            w = int(np.ceil(w / divisor)) * divisor
+
+        input_shape = (3, h, w)
+        result['resolution'] = input_shape
+
+        try:
+            cfg = Config.fromfile(config_file)
+            if hasattr(cfg, 'head_norm_cfg'):
+                cfg['head_norm_cfg'] = dict(type='SyncBN', requires_grad=True)
+                cfg['model']['roi_head']['bbox_head']['norm_cfg'] = dict(
+                    type='SyncBN', requires_grad=True)
+                cfg['model']['roi_head']['mask_head']['norm_cfg'] = dict(
+                    type='SyncBN', requires_grad=True)
+
+            if args.cfg_options is not None:
+                cfg.merge_from_dict(args.cfg_options)
+
+            model = MODELS.build(cfg.model)
+            input = torch.rand(1, *input_shape)
+            if torch.cuda.is_available():
+                model.cuda()
+                input = input.cuda()
+            model = revert_sync_batchnorm(model)
+            inputs = (input, )
+            model.eval()
+            outputs = get_model_complexity_info(
+                model, input_shape, inputs, show_table=False, show_arch=False)
+            flops = outputs['flops']
+            params = outputs['params']
+            activations = outputs['activations']
+            result['Get Types'] = 'direct'
+        except:  # noqa 772
+            logger = MMLogger.get_instance(name='MMLogger')
+            logger.warning(
+                'Direct get flops failed, try to get flops with data')
+            cfg = Config.fromfile(config_file)
+            if hasattr(cfg, 'head_norm_cfg'):
+                cfg['head_norm_cfg'] = dict(type='SyncBN', requires_grad=True)
+                cfg['model']['roi_head']['bbox_head']['norm_cfg'] = dict(
+                    type='SyncBN', requires_grad=True)
+                cfg['model']['roi_head']['mask_head']['norm_cfg'] = dict(
+                    type='SyncBN', requires_grad=True)
+            data_loader = Runner.build_dataloader(cfg.val_dataloader)
+            data_batch = next(iter(data_loader))
+            model = MODELS.build(cfg.model)
+            if torch.cuda.is_available():
+                model = model.cuda()
+            model = revert_sync_batchnorm(model)
+            model.eval()
+            _forward = model.forward
+            data = model.data_preprocessor(data_batch)
+            del data_loader
+            model.forward = partial(
+                _forward, data_samples=data['data_samples'])
+            outputs = get_model_complexity_info(
+                model,
+                input_shape,
+                data['inputs'],
+                show_table=False,
+                show_arch=False)
+            flops = outputs['flops']
+            params = outputs['params']
+            activations = outputs['activations']
+            result['Get Types'] = 'dataloader'
+
+        if args.flops_str:
+            flops = _format_size(flops)
+            params = _format_size(params)
+            activations = _format_size(activations)
+
+        result['flops'] = flops
+        result['params'] = params
+
+    return result
+
+
+def show_summary(summary_data, args):
+    table = Table(title='Validation Benchmark Regression Summary')
+    table.add_column('Model')
+    table.add_column('Validation')
+    table.add_column('Resolution (c, h, w)')
+    if args.flops:
+        table.add_column('Flops', justify='right', width=11)
+        table.add_column('Params', justify='right')
+
+    for model_name, summary in summary_data.items():
+        row = [model_name]
+        valid = summary['valid']
+        color = 'green' if valid == 'PASS' else 'red'
+        row.append(f'[{color}]{valid}[/{color}]')
+        if valid == 'PASS':
+            row.append(str(summary['resolution']))
+            if args.flops:
+                row.append(str(summary['flops']))
+                row.append(str(summary['params']))
+        table.add_row(*row)
+
+    console.print(table)
+    table_data = {
+        x.header: [Text.from_markup(y).plain for y in x.cells]
+        for x in table.columns
+    }
+    table_pd = pd.DataFrame(table_data)
+    table_pd.to_csv('./mmdetection_flops.csv')
+
+
+# Sample test whether the inference code is correct
+def main(args):
+    register_all_modules()
+    model_index_file = MMDET_ROOT / 'model-index.yml'
+    model_index = load(str(model_index_file))
+    model_index.build_models_with_collections()
+    models = OrderedDict({model.name: model for model in model_index.models})
+
+    logger = MMLogger(
+        'validation',
+        logger_name='validation',
+        log_file='benchmark_test_image.log',
+        log_level=logging.INFO)
+
+    if args.models:
+        patterns = [
+            re.compile(pattern.replace('+', '_')) for pattern in args.models
+        ]
+        filter_models = {}
+        for k, v in models.items():
+            k = k.replace('+', '_')
+            if any([re.match(pattern, k) for pattern in patterns]):
+                filter_models[k] = v
+        if len(filter_models) == 0:
+            print('No model found, please specify models in:')
+            print('\n'.join(models.keys()))
+            return
+        models = filter_models
+
+    summary_data = {}
+    tmpdir = tempfile.TemporaryDirectory()
+    for model_name, model_info in tqdm(models.items()):
+
+        if model_info.config is None:
+            continue
+
+        model_info.config = model_info.config.replace('%2B', '+')
+        config = Path(model_info.config)
+
+        try:
+            config.exists()
+        except:  # noqa 722
+            logger.error(f'{model_name}: {config} not found.')
+            continue
+
+        logger.info(f'Processing: {model_name}')
+
+        http_prefix = 'https://download.openmmlab.com/mmdetection/'
+        if args.checkpoint_root is not None:
+            root = args.checkpoint_root
+            if 's3://' in args.checkpoint_root:
+                from petrel_client.common.exception import AccessDeniedError
+                file_client = FileClient.infer_client(uri=root)
+                checkpoint = file_client.join_path(
+                    root, model_info.weights[len(http_prefix):])
+                try:
+                    exists = file_client.exists(checkpoint)
+                except AccessDeniedError:
+                    exists = False
+            else:
+                checkpoint = Path(root) / model_info.weights[len(http_prefix):]
+                exists = checkpoint.exists()
+            if exists:
+                checkpoint = str(checkpoint)
+            else:
+                print(f'WARNING: {model_name}: {checkpoint} not found.')
+                checkpoint = None
+        else:
+            checkpoint = None
+
+        try:
+            # build the model from a config file and a checkpoint file
+            result = inference(MMDET_ROOT / config, checkpoint, tmpdir.name,
+                               args, model_name)
+            result['valid'] = 'PASS'
+        except Exception:  # noqa 722
+            import traceback
+            logger.error(f'"{config}" :\n{traceback.format_exc()}')
+            result = {'valid': 'FAIL'}
+
+        summary_data[model_name] = result
+
+    tmpdir.cleanup()
+    show_summary(summary_data, args)
+
+
+if __name__ == '__main__':
+    args = parse_args()
+    main(args)

.dev_scripts/check_links.py

+# Modified from:
+# https://github.com/allenai/allennlp/blob/main/scripts/check_links.py
+
+import argparse
+import logging
+import os
+import pathlib
+import re
+import sys
+from multiprocessing.dummy import Pool
+from typing import NamedTuple, Optional, Tuple
+
+import requests
+from mmengine.logging import MMLogger
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(
+        description='Goes through all the inline-links '
+        'in markdown files and reports the breakages')
+    parser.add_argument(
+        '--num-threads',
+        type=int,
+        default=100,
+        help='Number of processes to confirm the link')
+    parser.add_argument('--https-proxy', type=str, help='https proxy')
+    parser.add_argument(
+        '--out',
+        type=str,
+        default='link_reports.txt',
+        help='output path of reports')
+    args = parser.parse_args()
+    return args
+
+
+OK_STATUS_CODES = (
+    200,
+    401,  # the resource exists but may require some sort of login.
+    403,  # ^ same
+    405,  # HEAD method not allowed.
+    # the resource exists, but our default 'Accept-' header may not
+    # match what the server can provide.
+    406,
+)
+
+
+class MatchTuple(NamedTuple):
+    source: str
+    name: str
+    link: str
+
+
+def check_link(
+        match_tuple: MatchTuple,
+        http_session: requests.Session,
+        logger: logging = None) -> Tuple[MatchTuple, bool, Optional[str]]:
+    reason: Optional[str] = None
+    if match_tuple.link.startswith('http'):
+        result_ok, reason = check_url(match_tuple, http_session)
+    else:
+        result_ok = check_path(match_tuple)
+    if logger is None:
+        print(f"  {'✓' if result_ok else '✗'} {match_tuple.link}")
+    else:
+        logger.info(f"  {'✓' if result_ok else '✗'} {match_tuple.link}")
+    return match_tuple, result_ok, reason
+
+
+def check_url(match_tuple: MatchTuple,
+              http_session: requests.Session) -> Tuple[bool, str]:
+    """Check if a URL is reachable."""
+    try:
+        result = http_session.head(
+            match_tuple.link, timeout=5, allow_redirects=True)
+        return (
+            result.ok or result.status_code in OK_STATUS_CODES,
+            f'status code = {result.status_code}',
+        )
+    except (requests.ConnectionError, requests.Timeout):
+        return False, 'connection error'
+
+
+def check_path(match_tuple: MatchTuple) -> bool:
+    """Check if a file in this repository exists."""
+    relative_path = match_tuple.link.split('#')[0]
+    full_path = os.path.join(
+        os.path.dirname(str(match_tuple.source)), relative_path)
+    return os.path.exists(full_path)
+
+
+def main():
+    args = parse_args()
+
+    # setup logger
+    logger = MMLogger.get_instance(name='mmdet', log_file=args.out)
+
+    # setup https_proxy
+    if args.https_proxy:
+        os.environ['https_proxy'] = args.https_proxy
+
+    # setup http_session
+    http_session = requests.Session()
+    for resource_prefix in ('http://', 'https://'):
+        http_session.mount(
+            resource_prefix,
+            requests.adapters.HTTPAdapter(
+                max_retries=5,
+                pool_connections=20,
+                pool_maxsize=args.num_threads),
+        )
+
+    logger.info('Finding all markdown files in the current directory...')
+
+    project_root = (pathlib.Path(__file__).parent / '..').resolve()
+    markdown_files = project_root.glob('**/*.md')
+
+    all_matches = set()
+    url_regex = re.compile(r'\[([^!][^\]]+)\]\(([^)(]+)\)')
+    for markdown_file in markdown_files:
+        with open(markdown_file) as handle:
+            for line in handle.readlines():
+                matches = url_regex.findall(line)
+                for name, link in matches:
+                    if 'localhost' not in link:
+                        all_matches.add(
+                            MatchTuple(
+                                source=str(markdown_file),
+                                name=name,
+                                link=link))
+
+    logger.info(f'  {len(all_matches)} markdown files found')
+    logger.info('Checking to make sure we can retrieve each link...')
+
+    with Pool(processes=args.num_threads) as pool:
+        results = pool.starmap(check_link, [(match, http_session, logger)
+                                            for match in list(all_matches)])
+
+    # collect unreachable results
+    unreachable_results = [(match_tuple, reason)
+                           for match_tuple, success, reason in results
+                           if not success]
+
+    if unreachable_results:
+        logger.info('================================================')
+        logger.info(f'Unreachable links ({len(unreachable_results)}):')
+        for match_tuple, reason in unreachable_results:
+            logger.info('  > Source: ' + match_tuple.source)
+            logger.info('    Name: ' + match_tuple.name)
+            logger.info('    Link: ' + match_tuple.link)
+            if reason is not None:
+                logger.info('    Reason: ' + reason)
+        sys.exit(1)
+    logger.info('No Unreachable link found.')
+
+
+if __name__ == '__main__':
+    main()

.dev_scripts/convert_test_benchmark_script.py

+# Copyright (c) OpenMMLab. All rights reserved.
+import argparse
+import os
+import os.path as osp
+
+from mmengine import Config
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(
+        description='Convert benchmark model list to script')
+    parser.add_argument('config', help='test config file path')
+    parser.add_argument('--port', type=int, default=29666, help='dist port')
+    parser.add_argument(
+        '--run', action='store_true', help='run script directly')
+    parser.add_argument(
+        '--out', type=str, help='path to save model benchmark script')
+
+    args = parser.parse_args()
+    return args
+
+
+def process_model_info(model_info, work_dir):
+    config = model_info['config'].strip()
+    fname, _ = osp.splitext(osp.basename(config))
+    job_name = fname
+    work_dir = '$WORK_DIR/' + fname
+    checkpoint = model_info['checkpoint'].strip()
+    return dict(
+        config=config,
+        job_name=job_name,
+        work_dir=work_dir,
+        checkpoint=checkpoint)
+
+
+def create_test_bash_info(commands, model_test_dict, port, script_name,
+                          partition):
+    config = model_test_dict['config']
+    job_name = model_test_dict['job_name']
+    checkpoint = model_test_dict['checkpoint']
+    work_dir = model_test_dict['work_dir']
+
+    echo_info = f' \necho \'{config}\' &'
+    commands.append(echo_info)
+    commands.append('\n')
+
+    command_info = f'GPUS=8  GPUS_PER_NODE=8  ' \
+                   f'CPUS_PER_TASK=$CPUS_PRE_TASK {script_name} '
+
+    command_info += f'{partition} '
+    command_info += f'{job_name} '
+    command_info += f'{config} '
+    command_info += f'$CHECKPOINT_DIR/{checkpoint} '
+    command_info += f'--work-dir {work_dir} '
+
+    command_info += f'--cfg-option env_cfg.dist_cfg.port={port} '
+    command_info += ' &'
+
+    commands.append(command_info)
+
+
+def main():
+    args = parse_args()
+    if args.out:
+        out_suffix = args.out.split('.')[-1]
+        assert args.out.endswith('.sh'), \
+            f'Expected out file path suffix is .sh, but get .{out_suffix}'
+    assert args.out or args.run, \
+        ('Please specify at least one operation (save/run/ the '
+         'script) with the argument "--out" or "--run"')
+
+    commands = []
+    partition_name = 'PARTITION=$1 '
+    commands.append(partition_name)
+    commands.append('\n')
+
+    checkpoint_root = 'CHECKPOINT_DIR=$2 '
+    commands.append(checkpoint_root)
+    commands.append('\n')
+
+    work_dir = 'WORK_DIR=$3 '
+    commands.append(work_dir)
+    commands.append('\n')
+
+    cpus_pre_task = 'CPUS_PER_TASK=${4:-2} '
+    commands.append(cpus_pre_task)
+    commands.append('\n')
+
+    script_name = osp.join('tools', 'slurm_test.sh')
+    port = args.port
+
+    cfg = Config.fromfile(args.config)
+
+    for model_key in cfg:
+        model_infos = cfg[model_key]
+        if not isinstance(model_infos, list):
+            model_infos = [model_infos]
+        for model_info in model_infos:
+            print('processing: ', model_info['config'])
+            model_test_dict = process_model_info(model_info, work_dir)
+            create_test_bash_info(commands, model_test_dict, port, script_name,
+                                  '$PARTITION')
+            port += 1
+
+    command_str = ''.join(commands)
+    if args.out:
+        with open(args.out, 'w') as f:
+            f.write(command_str)
+    if args.run:
+        os.system(command_str)
+
+
+if __name__ == '__main__':
+    main()

.dev_scripts/convert_train_benchmark_script.py

+# Copyright (c) OpenMMLab. All rights reserved.
+import argparse
+import os
+import os.path as osp
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(
+        description='Convert benchmark model json to script')
+    parser.add_argument(
+        'txt_path', type=str, help='txt path output by benchmark_filter')
+    parser.add_argument(
+        '--run', action='store_true', help='run script directly')
+    parser.add_argument(
+        '--out', type=str, help='path to save model benchmark script')
+
+    args = parser.parse_args()
+    return args
+
+
+def determine_gpus(cfg_name):
+    gpus = 8
+    gpus_pre_node = 8
+
+    if cfg_name.find('16x') >= 0:
+        gpus = 16
+    elif cfg_name.find('4xb4') >= 0:
+        gpus = 4
+        gpus_pre_node = 4
+    elif 'lad' in cfg_name:
+        gpus = 2
+        gpus_pre_node = 2
+
+    return gpus, gpus_pre_node
+
+
+def main():
+    args = parse_args()
+    if args.out:
+        out_suffix = args.out.split('.')[-1]
+        assert args.out.endswith('.sh'), \
+            f'Expected out file path suffix is .sh, but get .{out_suffix}'
+    assert args.out or args.run, \
+        ('Please specify at least one operation (save/run/ the '
+         'script) with the argument "--out" or "--run"')
+
+    root_name = './tools'
+    train_script_name = osp.join(root_name, 'slurm_train.sh')
+
+    commands = []
+    partition_name = 'PARTITION=$1 '
+    commands.append(partition_name)
+    commands.append('\n')
+
+    work_dir = 'WORK_DIR=$2 '
+    commands.append(work_dir)
+    commands.append('\n')
+
+    cpus_pre_task = 'CPUS_PER_TASK=${3:-4} '
+    commands.append(cpus_pre_task)
+    commands.append('\n')
+    commands.append('\n')
+
+    with open(args.txt_path, 'r') as f:
+        model_cfgs = f.readlines()
+        for i, cfg in enumerate(model_cfgs):
+            cfg = cfg.strip()
+            if len(cfg) == 0:
+                continue
+            # print cfg name
+            echo_info = f'echo \'{cfg}\' &'
+            commands.append(echo_info)
+            commands.append('\n')
+
+            fname, _ = osp.splitext(osp.basename(cfg))
+            out_fname = '$WORK_DIR/' + fname
+
+            gpus, gpus_pre_node = determine_gpus(cfg)
+            command_info = f'GPUS={gpus}  GPUS_PER_NODE={gpus_pre_node}  ' \
+                           f'CPUS_PER_TASK=$CPUS_PRE_TASK {train_script_name} '
+            command_info += '$PARTITION '
+            command_info += f'{fname} '
+            command_info += f'{cfg} '
+            command_info += f'{out_fname} '
+
+            command_info += '--cfg-options default_hooks.checkpoint.' \
+                            'max_keep_ckpts=1 '
+            command_info += '&'
+
+            commands.append(command_info)
+
+            if i < len(model_cfgs):
+                commands.append('\n')
+
+        command_str = ''.join(commands)
+        if args.out:
+            with open(args.out, 'w') as f:
+                f.write(command_str)
+        if args.run:
+            os.system(command_str)
+
+
+if __name__ == '__main__':
+    main()

.dev_scripts/covignore.cfg

+# Each line should be the relative path to the root directory
+# of this repo. Support regular expression as well.
+# For example:
+
+.*/__init__.py

.dev_scripts/diff_coverage_test.sh

+#!/bin/bash
+
+readarray -t IGNORED_FILES < $( dirname "$0" )/covignore.cfg
+REUSE_COVERAGE_REPORT=${REUSE_COVERAGE_REPORT:-0}
+REPO=${1:-"origin"}
+BRANCH=${2:-"refactor_dev"}
+
+git fetch $REPO $BRANCH
+
+PY_FILES=""
+for FILE_NAME in $(git diff --name-only ${REPO}/${BRANCH}); do
+    # Only test python files in mmdet/ existing in current branch, and not ignored in covignore.cfg
+    if [ ${FILE_NAME: -3} == ".py" ] && [ ${FILE_NAME:0:6} == "mmdet/" ] && [ -f "$FILE_NAME" ]; then
+        IGNORED=false
+        for IGNORED_FILE_NAME in "${IGNORED_FILES[@]}"; do
+            # Skip blank lines
+            if [ -z "$IGNORED_FILE_NAME" ]; then
+                continue
+            fi
+            if [ "${IGNORED_FILE_NAME::1}" != "#" ] && [[ "$FILE_NAME" =~ $IGNORED_FILE_NAME ]]; then
+                echo "Ignoring $FILE_NAME"
+                IGNORED=true
+                break
+            fi
+        done
+        if [ "$IGNORED" = false ]; then
+            PY_FILES="$PY_FILES $FILE_NAME"
+        fi
+    fi
+done
+
+# Only test the coverage when PY_FILES are not empty, otherwise they will test the entire project
+if [ ! -z "${PY_FILES}" ]
+then
+    if [ "$REUSE_COVERAGE_REPORT" == "0" ]; then
+        coverage run --branch --source mmdet -m pytest tests/
+    fi
+    coverage report --fail-under 80 -m $PY_FILES
+    interrogate -v --ignore-init-method --ignore-module --ignore-nested-functions --ignore-magic --ignore-regex "__repr__" --fail-under 95 $PY_FILES
+fi

.dev_scripts/download_checkpoints.py

+# Copyright (c) OpenMMLab. All rights reserved.
+
+import argparse
+import math
+import os
+import os.path as osp
+from multiprocessing import Pool
+
+import torch
+from mmengine.config import Config
+from mmengine.utils import mkdir_or_exist
+
+
+def download(url, out_file, min_bytes=math.pow(1024, 2), progress=True):
+    # math.pow(1024, 2) is mean 1 MB
+    assert_msg = f"Downloaded url '{url}' does not exist " \
+                 f'or size is < min_bytes={min_bytes}'
+    try:
+        print(f'Downloading {url} to {out_file}...')
+        torch.hub.download_url_to_file(url, str(out_file), progress=progress)
+        assert osp.exists(
+            out_file) and osp.getsize(out_file) > min_bytes, assert_msg
+    except Exception as e:
+        if osp.exists(out_file):
+            os.remove(out_file)
+        print(f'ERROR: {e}\nRe-attempting {url} to {out_file} ...')
+        os.system(f"curl -L '{url}' -o '{out_file}' --retry 3 -C -"
+                  )  # curl download, retry and resume on fail
+    finally:
+        if osp.exists(out_file) and osp.getsize(out_file) < min_bytes:
+            os.remove(out_file)  # remove partial downloads
+
+        if not osp.exists(out_file):
+            print(f'ERROR: {assert_msg}\n')
+        print('=========================================\n')
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(description='Download checkpoints')
+    parser.add_argument('config', help='test config file path')
+    parser.add_argument(
+        'out', type=str, help='output dir of checkpoints to be stored')
+    parser.add_argument(
+        '--nproc', type=int, default=16, help='num of Processes')
+    parser.add_argument(
+        '--intranet',
+        action='store_true',
+        help='switch to internal network url')
+    args = parser.parse_args()
+    return args
+
+
+if __name__ == '__main__':
+    args = parse_args()
+    mkdir_or_exist(args.out)
+
+    cfg = Config.fromfile(args.config)
+
+    checkpoint_url_list = []
+    checkpoint_out_list = []
+
+    for model in cfg:
+        model_infos = cfg[model]
+        if not isinstance(model_infos, list):
+            model_infos = [model_infos]
+        for model_info in model_infos:
+            checkpoint = model_info['checkpoint']
+            out_file = osp.join(args.out, checkpoint)
+            if not osp.exists(out_file):
+
+                url = model_info['url']
+                if args.intranet is True:
+                    url = url.replace('.com', '.sensetime.com')
+                    url = url.replace('https', 'http')
+
+                checkpoint_url_list.append(url)
+                checkpoint_out_list.append(out_file)
+
+    if len(checkpoint_url_list) > 0:
+        pool = Pool(min(os.cpu_count(), args.nproc))
+        pool.starmap(download, zip(checkpoint_url_list, checkpoint_out_list))
+    else:
+        print('No files to download!')

.dev_scripts/gather_models.py

+# Copyright (c) OpenMMLab. All rights reserved.
+import argparse
+import glob
+import os
+import os.path as osp
+import shutil
+import subprocess
+import time
+from collections import OrderedDict
+
+import torch
+import yaml
+from mmengine.config import Config
+from mmengine.fileio import dump
+from mmengine.utils import digit_version, mkdir_or_exist, scandir
+
+
+def ordered_yaml_dump(data, stream=None, Dumper=yaml.SafeDumper, **kwds):
+
+    class OrderedDumper(Dumper):
+        pass
+
+    def _dict_representer(dumper, data):
+        return dumper.represent_mapping(
+            yaml.resolver.BaseResolver.DEFAULT_MAPPING_TAG, data.items())
+
+    OrderedDumper.add_representer(OrderedDict, _dict_representer)
+    return yaml.dump(data, stream, OrderedDumper, **kwds)
+
+
+def process_checkpoint(in_file, out_file):
+    checkpoint = torch.load(in_file, map_location='cpu')
+    # remove optimizer for smaller file size
+    if 'optimizer' in checkpoint:
+        del checkpoint['optimizer']
+    if 'ema_state_dict' in checkpoint:
+        del checkpoint['ema_state_dict']
+
+    # remove ema state_dict
+    for key in list(checkpoint['state_dict']):
+        if key.startswith('ema_'):
+            checkpoint['state_dict'].pop(key)
+        elif key.startswith('data_preprocessor'):
+            checkpoint['state_dict'].pop(key)
+
+    # if it is necessary to remove some sensitive data in checkpoint['meta'],
+    # add the code here.
+    if digit_version(torch.__version__) >= digit_version('1.6'):
+        torch.save(checkpoint, out_file, _use_new_zipfile_serialization=False)
+    else:
+        torch.save(checkpoint, out_file)
+    sha = subprocess.check_output(['sha256sum', out_file]).decode()
+    final_file = out_file.rstrip('.pth') + '-{}.pth'.format(sha[:8])
+    subprocess.Popen(['mv', out_file, final_file])
+    return final_file
+
+
+def is_by_epoch(config):
+    cfg = Config.fromfile('./configs/' + config)
+    return cfg.train_cfg.type == 'EpochBasedTrainLoop'
+
+
+def get_final_epoch_or_iter(config):
+    cfg = Config.fromfile('./configs/' + config)
+    if cfg.train_cfg.type == 'EpochBasedTrainLoop':
+        return cfg.train_cfg.max_epochs
+    else:
+        return cfg.train_cfg.max_iters
+
+
+def get_best_epoch_or_iter(exp_dir):
+    best_epoch_iter_full_path = list(
+        sorted(glob.glob(osp.join(exp_dir, 'best_*.pth'))))[-1]
+    best_epoch_or_iter_model_path = best_epoch_iter_full_path.split('/')[-1]
+    best_epoch_or_iter = best_epoch_or_iter_model_path.\
+        split('_')[-1].split('.')[0]
+    return best_epoch_or_iter_model_path, int(best_epoch_or_iter)
+
+
+def get_real_epoch_or_iter(config):
+    cfg = Config.fromfile('./configs/' + config)
+    if cfg.train_cfg.type == 'EpochBasedTrainLoop':
+        epoch = cfg.train_cfg.max_epochs
+        return epoch
+    else:
+        return cfg.train_cfg.max_iters
+
+
+def get_final_results(log_json_path,
+                      epoch_or_iter,
+                      results_lut='coco/bbox_mAP',
+                      by_epoch=True):
+    result_dict = dict()
+    with open(log_json_path) as f:
+        r = f.readlines()[-1]
+        last_metric = r.split(',')[0].split(': ')[-1].strip()
+    result_dict[results_lut] = last_metric
+    return result_dict
+
+
+def get_dataset_name(config):
+    # If there are more dataset, add here.
+    name_map = dict(
+        CityscapesDataset='Cityscapes',
+        CocoDataset='COCO',
+        CocoPanopticDataset='COCO',
+        DeepFashionDataset='Deep Fashion',
+        LVISV05Dataset='LVIS v0.5',
+        LVISV1Dataset='LVIS v1',
+        VOCDataset='Pascal VOC',
+        WIDERFaceDataset='WIDER Face',
+        OpenImagesDataset='OpenImagesDataset',
+        OpenImagesChallengeDataset='OpenImagesChallengeDataset',
+        Objects365V1Dataset='Objects365 v1',
+        Objects365V2Dataset='Objects365 v2')
+    cfg = Config.fromfile('./configs/' + config)
+    return name_map[cfg.dataset_type]
+
+
+def find_last_dir(model_dir):
+    dst_times = []
+    for time_stamp in os.scandir(model_dir):
+        if osp.isdir(time_stamp):
+            dst_time = time.mktime(
+                time.strptime(time_stamp.name, '%Y%m%d_%H%M%S'))
+            dst_times.append([dst_time, time_stamp.name])
+    return max(dst_times, key=lambda x: x[0])[1]
+
+
+def convert_model_info_to_pwc(model_infos):
+    pwc_files = {}
+    for model in model_infos:
+        cfg_folder_name = osp.split(model['config'])[-2]
+        pwc_model_info = OrderedDict()
+        pwc_model_info['Name'] = osp.split(model['config'])[-1].split('.')[0]
+        pwc_model_info['In Collection'] = 'Please fill in Collection name'
+        pwc_model_info['Config'] = osp.join('configs', model['config'])
+
+        # get metadata
+        meta_data = OrderedDict()
+        if 'epochs' in model:
+            meta_data['Epochs'] = get_real_epoch_or_iter(model['config'])
+        else:
+            meta_data['Iterations'] = get_real_epoch_or_iter(model['config'])
+        pwc_model_info['Metadata'] = meta_data
+
+        # get dataset name
+        dataset_name = get_dataset_name(model['config'])
+
+        # get results
+        results = []
+        # if there are more metrics, add here.
+        if 'bbox_mAP' in model['results']:
+            metric = round(model['results']['bbox_mAP'] * 100, 1)
+            results.append(
+                OrderedDict(
+                    Task='Object Detection',
+                    Dataset=dataset_name,
+                    Metrics={'box AP': metric}))
+        if 'segm_mAP' in model['results']:
+            metric = round(model['results']['segm_mAP'] * 100, 1)
+            results.append(
+                OrderedDict(
+                    Task='Instance Segmentation',
+                    Dataset=dataset_name,
+                    Metrics={'mask AP': metric}))
+        if 'PQ' in model['results']:
+            metric = round(model['results']['PQ'], 1)
+            results.append(
+                OrderedDict(
+                    Task='Panoptic Segmentation',
+                    Dataset=dataset_name,
+                    Metrics={'PQ': metric}))
+        pwc_model_info['Results'] = results
+
+        link_string = 'https://download.openmmlab.com/mmdetection/v3.0/'
+        link_string += '{}/{}'.format(model['config'].rstrip('.py'),
+                                      osp.split(model['model_path'])[-1])
+        pwc_model_info['Weights'] = link_string
+        if cfg_folder_name in pwc_files:
+            pwc_files[cfg_folder_name].append(pwc_model_info)
+        else:
+            pwc_files[cfg_folder_name] = [pwc_model_info]
+    return pwc_files
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(description='Gather benchmarked models')
+    parser.add_argument(
+        'root',
+        type=str,
+        default='work_dirs',
+        help='root path of benchmarked models to be gathered')
+    parser.add_argument(
+        '--out',
+        type=str,
+        default='gather',
+        help='output path of gathered models to be stored')
+    parser.add_argument(
+        '--best',
+        action='store_true',
+        help='whether to gather the best model.')
+
+    args = parser.parse_args()
+    return args
+
+
+def main():
+    args = parse_args()
+    models_root = args.root
+    models_out = args.out
+    mkdir_or_exist(models_out)
+
+    # find all models in the root directory to be gathered
+    raw_configs = list(scandir('./configs', '.py', recursive=True))
+
+    # filter configs that is not trained in the experiments dir
+    used_configs = []
+    for raw_config in raw_configs:
+        if osp.exists(osp.join(models_root, raw_config)):
+            used_configs.append(raw_config)
+    print(f'Find {len(used_configs)} models to be gathered')
+
+    # find final_ckpt and log file for trained each config
+    # and parse the best performance
+    model_infos = []
+    for used_config in used_configs:
+        exp_dir = osp.join(models_root, used_config)
+        by_epoch = is_by_epoch(used_config)
+        # check whether the exps is finished
+        if args.best is True:
+            final_model, final_epoch_or_iter = get_best_epoch_or_iter(exp_dir)
+        else:
+            final_epoch_or_iter = get_final_epoch_or_iter(used_config)
+            final_model = '{}_{}.pth'.format('epoch' if by_epoch else 'iter',
+                                             final_epoch_or_iter)
+
+        model_path = osp.join(exp_dir, final_model)
+        # skip if the model is still training
+        if not osp.exists(model_path):
+            continue
+
+        # get the latest logs
+        latest_exp_name = find_last_dir(exp_dir)
+        latest_exp_json = osp.join(exp_dir, latest_exp_name, 'vis_data',
+                                   latest_exp_name + '.json')
+
+        model_performance = get_final_results(
+            latest_exp_json, final_epoch_or_iter, by_epoch=by_epoch)
+
+        if model_performance is None:
+            continue
+
+        model_info = dict(
+            config=used_config,
+            results=model_performance,
+            final_model=final_model,
+            latest_exp_json=latest_exp_json,
+            latest_exp_name=latest_exp_name)
+        model_info['epochs' if by_epoch else 'iterations'] =\
+            final_epoch_or_iter
+        model_infos.append(model_info)
+
+    # publish model for each checkpoint
+    publish_model_infos = []
+    for model in model_infos:
+        model_publish_dir = osp.join(models_out, model['config'].rstrip('.py'))
+        mkdir_or_exist(model_publish_dir)
+
+        model_name = osp.split(model['config'])[-1].split('.')[0]
+
+        model_name += '_' + model['latest_exp_name']
+        publish_model_path = osp.join(model_publish_dir, model_name)
+        trained_model_path = osp.join(models_root, model['config'],
+                                      model['final_model'])
+
+        # convert model
+        final_model_path = process_checkpoint(trained_model_path,
+                                              publish_model_path)
+
+        # copy log
+        shutil.copy(model['latest_exp_json'],
+                    osp.join(model_publish_dir, f'{model_name}.log.json'))
+
+        # copy config to guarantee reproducibility
+        config_path = model['config']
+        config_path = osp.join(
+            'configs',
+            config_path) if 'configs' not in config_path else config_path
+        target_config_path = osp.split(config_path)[-1]
+        shutil.copy(config_path, osp.join(model_publish_dir,
+                                          target_config_path))
+
+        model['model_path'] = final_model_path
+        publish_model_infos.append(model)
+
+    models = dict(models=publish_model_infos)
+    print(f'Totally gathered {len(publish_model_infos)} models')
+    dump(models, osp.join(models_out, 'model_info.json'))
+
+    pwc_files = convert_model_info_to_pwc(publish_model_infos)
+    for name in pwc_files:
+        with open(osp.join(models_out, name + '_metafile.yml'), 'w') as f:
+            ordered_yaml_dump(pwc_files[name], f, encoding='utf-8')
+
+
+if __name__ == '__main__':
+    main()